Positively Selected Effector Genes and Their Contribution to Virulence in the Smut Fungus Sporisorium reilianum

Positively Selected Effector Genes and Their Contribution to Virulence in the Smut Fungus... Plants and fungi display a broad range of interactions in natural and agricultural ecosystems ranging from symbiosis to parasitism. These ecological interactions result in coevolution between genes belonging to different partners. A well- understood example is secreted fungal effector proteins and their host targets, which play an important role in pathogenic interactions. Biotrophic smut fungi (Basidiomycota) are well-suited to investigate the evolution of plant pathogens, because several reference genomes and genetic tools are available for these species. Here, we used the genomes of Sporisorium reilianum f. sp. zeae and S. reilianum f. sp. reilianum, two closely related formae speciales infecting maize and sorghum, respectively, together with the genomes of Ustilago hordei, Ustilago maydis,and Sporisorium scitamineum to identify and characterize genes displaying signatures of positive selection. We identified 154 gene families having under- gone positive selection during species divergence in at least one lineage, among which 77% were identified in the two investigated formae speciales of S. reilianum. Remarkably, only 29% of positively selected genes encode predicted secreted proteins. We assessed the contribution to virulence of nine of these candidate effector genes in S. reilianum f. sp. zeae by deleting individual genes, including a homologue of the effector gene pit2 previously characterized in U. maydis.Only the pit2 deletion mutant was found to be strongly reduced in virulence. Additional experiments are required to understand the molecular mechanisms underlying the selection forces acting on the other candidate effector genes, as well as the large fraction of positively selected genes encoding predicted cytoplasmic proteins. Key words: positive selection, effector evolution, smut fungi, comparative genomics, virulence. Introduction and fossil records (Remy et al. 1994; Gehrig et al. 1996; Plants and fungi have a long history of coevolution since the Martin et al. 2017). Different forms of plant–fungus interac- emergence of pioneering land plants 400 Ma. The develop- tions have evolved, including mutualistic symbiosis where ment of early plants was likely supported by associations with both plant and fungus benefit (Parniske 2008), and patho- symbiotic fungi, as suggested by analyses of ribosomal RNAs genic interactions where fungal colonization greatly reduces The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Genome Biol. Evol. 10(2):629–645. doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 629 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE plant fitness (Dean et al. 2012). Pathogenic interactions play Caryophyllaceae species (Aguileta et al. 2010), the wheat critical roles in natural and agricultural ecosystems, and un- pathogen Zymoseptoria tritici (Stukenbrock et al. 2011), the derstanding the evolutionary mechanisms shaping them is of rust fungus Melampsora larici-populina (Hacquard et al. great importance to plant production, food security, and pro- 2012), the rice BLAST fungus Magnaporthe oryzae (Huang tection of biodiversity in natural ecosystems (Fisher et al. et al. 2014), the wheat stem rust fungus Puccinia graminis 2012; Bagchi et al. 2014). f. sp. tritici (Sperschneider et al. 2014), the Irish potato famine Secreted fungal effector proteins are key players in patho- pathogen Phytophthora infestans (Dong et al. 2014), a group genic interactions as they are involved in protecting and of Fusarium species (Sperschneider et al. 2015), and a group shielding growing hyphae, suppressing plant defense of smut fungi parasitizing different grasses and a dicot host responses and changing plant physiology to support growth (Sharma et al. 2014, 2015). Yet, as only a few genes under of the pathogen (Stergiopoulos and de Wit 2009; de Jonge positive selection have been functionally studied, the link be- et al. 2011; Giraldo and Valent 2013). Many effector proteins tween the selected genotypes and their corresponding phe- lack known functional domains, and expression of a subset of notypes are only beginning to be understood and only a few effectors is linked to plant colonization (Lo Presti et al. 2015; studies used evolutionary predictions to unravel the molecular Toruno et al. 2016; Franceschetti et al. 2017; Lanver et al. mechanisms of host adaptation. For example, the population 2017). Effector proteins with a strong effect on virulence phe- genomics study in the wheat pathogen Z. tritici which identi- notype are thought to coevolve with their plant targets either fied candidate effector genes under positive selection in an arms race or a trench-warfare scenario (Brown and (Stukenbrock et al. 2011) was followed up experimentally, Tellier 2011). In the former, fungal effectors manipulating and in this case it was shown that the deletion of three of the host are under positive directional selection, and plant four candidate genes reduced virulence (Poppe et al. 2015). In targets evolve in response to changes in effector proteins the gray mold fungus Botrytis cinerea, four positively selected (Rovenich et al. 2014). In the latter scenario, sets of alleles genes were deleted without affecting virulence, and this find- are maintained by balancing selection in both host and path- ing was attributed to functional redundancy, the limited num- ogen populations (Brown and Tellier 2011; Tellier et al. 2014). ber of tested host plants, or experimental conditions different Several methods are available for identifying genomic regions from natural infections (Aguileta 2012). A study of the oomy- under selection (Nielsen 2005; Aguileta et al. 2009, 2010). It cete effector protein EpiC1 showed that a single amino acid has been proposed that genes with signatures of positive se- substitution at a site under positive selection affected the lection have important functions during host pathogen inter- binding affinity of different host proteases determining host action or have contributed to host specialization (Tiffin and specificity (Dong et al. 2014). Moeller 2006). It is therefore expected that the deletion of Smut fungi, belonging to the division of Basidiomycota, are such genes reduces virulence when tested on a susceptible a group of about 550 species parasitizing mostly grasses, in- host. cluding important crops like maize, sorghum, oat, barley, and Depending on the aim of the investigation, studies identi- sugarcane (Begerow et al. 2014). In smut fungi, sexual repro- fying genes with signatures of positive selection are carried duction is linked to pathogenic development and smut fungi out within or between species. Although studies on the pop- therefore depend on successful plant colonization to com- ulation level focus on recent and ongoing selective processes plete their life cycle. As biotrophic pathogens, they require and are instrumental in the understanding of adaptation, living plant tissue for establishing a successful interaction comparative genomic studies employing different species en- (Martinez-Espinoza et al. 2002). With few exceptions like compass a broader time span and provide insight into the Ustilago maydis, smut fungi usually develop symptoms only underlying genetic basis of host specialization (Plissonneau in the female or male inflorescence of their respective host et al. 2017). The signature of positive selection in such case plants. During the last 10 years, quality draft genome sequen- typically takes the form of an excess of divergence between ces of prominent species were obtained, including U. maydis, species due to increased fixation of mutations by selective thecausative agentofsmutdisease on maize and teosinte sweeps compared with a neutral expectation (Yang and (Kamper et al. 2006), Sporisorium reilianum causing head Nielsen 1998). This is commonly measured by the ratio of smut of maize and sorghum (Schirawski et al. 2010), nonsynonymous (d ) over synonymous (d ) divergence and Ustilago hordei infecting barley (Laurie et al. 2012), and N S a d /d ratio> 1 is taken as evidence for positive selection Sporisorium scitamineum parasitizing sugarcane (Queetal. N S under the assumption that synonymous substitutions are neu- 2014; Taniguti et al. 2015; Dutheil et al. 2016). The head tral while nonsynonymous are not. Positive selection studies in smut fungus S. reilianum occurs in two formae speciales a number of plant pathogen systems revealed that genes that infect maize (S. reilianum f. sp. zeae) or sorghum (S. encoding secreted effector proteins are enriched in signatures reilianum f. sp. reilianum)(Zuther et al. 2012). The concept of positive selection (Mo ¨ ller and Stukenbrock 2017). Such of formae speciales is used in phytopathology to distinguish studies include investigations in diverse plant pathogens like members of the same species based on their ability to colonize Microbotryum species causing anther-smut disease of a certain host plant (in this example maize or sorghum) 630 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE (Anikster 1984). The divergence of U. hordei, U. maydis, S. two members were aligned on the codon level using MACSE scitamineum,and S. reilianum wasinferredtohave occurred 1.01 b (Ranwez et al. 2011) and on the protein level using in the interval of seven and 50 Ma (Munkacsi et al. 2007). The PRANK v.100802 (Lo ¨ ytynoja and Goldman 2008). The result- availability of genome sequences of several species with dif- ing alignments were subsequently compared and column ferent host ranges, together with established tools for genetic scores (CS) computed for each position in the alignment manipulations (Brachmann et al. 2004; Kamper 2004; (Thompson et al. 1999). Only positions with CS of 100% Khrunyk et al. 2010; Schuster et al. 2016) make this group (that is, alignment columns identically found by both meth- of smut fungi particularly interesting to study the evolution of ods) and a maximum of 30% gaps were retained for further effector genes as well as their contributions to virulence, spe- analysis. ciation, and host specificity. Here, we employed the genome of the recently sequenced strain S. reilianum f. sp. reilianum SRS1_H2-8 (http://www.ebi. Estimation of Genome-Wide Divergence Values and ac.uk/ena/data/view/LT795054-LT795076)(Zuther et al. Divergence Times 2012) together with the genomes of U. maydis, U. hordei, S. scitamineum,and S. reilianum f. sp. zeae to identify poten- The five genomes of U. hordei, U. maydis, S. scitamineum, tial effector genes with signatures of positive selection. S. reilianum f. sp. zeae,and S. reilianum f. sp. reilianum Candidate genes were individually deleted in S. reilianum f. were aligned using the Multiz genome aligner from the sp. zeae and the phenotype of the deletion strains was TBA package (Blanchette et al. 2004) and projected on the assessed after infection of maize in order to understand their U. maydis genome as reference. The resulting multiple function with respect to virulence. We report that the deletion genome alignment had a total size of 21 Mb and was of one candidate gene, pit2, led to a strong reduction in vir- further restricted to regions with homologous sequences ulence and we further discuss hypotheses on the origin of in the five species (total length after this step: 14.3 Mb) positive selection for the other candidate genes. and processed to remove coding regions. The final non- coding sequence alignment had a total length of 2.2 Mb, for which pairwise nucleotide similarities were computed in nonoverlapping windows of 10 kb. Materials and Methods Gene families with exactly one member in each species Construction of Homologous Protein Families were concatenated and pairwise protein sequence similarities Fungal species used in this study, their number of gene mod- computed using the seqinr package for R (Charif and Lobry els, number of predicted secreted proteins, and sources of 2007). Protein alignments were also used to infer dates of genome data are listed in supplementary table S1, divergence, using a relaxed clock model. The PhyloBayes ver- Supplementary Material online. The predicted proteome of sion 4.1 software (Lartillot et al. 2009) was used with the the five smut fungi U. hordei, Umaydis, S. scitamineum, S. auto-correlated model of Thorne et al. (Thorne et al. 1998) reilianum f. sp. zeae,and S. reilianum f. sp. reilianum were under a GTRþ CAT model. A unique calibration point was used to perform an all-against-all BLASTp search (Altschul used, based on the divergence time of the most divergent et al. 1990). The SiLiX algorithm was subsequently used to lineage U. hordei, previously estimated to have occurred be- infer homology relationships based on the BLAST hits (Miele tween 27 and 21 Myr (Bakkeren and Kronstad 2007). A uni- et al. 2011). Two parameters are considered to decide form prior was used on this interval for the Monte-Carlo whether a BLAST hit can be taken as evidence for homology: Markov Chain. As convergence issues arise when large align- the percent identity between two sequences and the relative ments (>20,000 positions) are used, we followed the length of the hit compared with the total length of the two PhyloBayes authors’ recommendation to conduct a jackknife sequences, hereby referred to as “coverage.” In order to max- procedure. We generated three data sets of circa 20,000 imize the number of families comprising 1:1 orthologues (that amino acids by randomly sampling families and concatenating is families that have an equal number of members in each the corresponding alignments. Two chains were run in each species), SiLiX (Miele et al. 2011) was runwitha range for case and convergence was assessed. Sampling was per- coverage and identity thresholds between 5% and 95% in formed after a burning of 10,000 iterations, and every ten 5% steps. An identity of 40% and coverage between 5% and subsequent iterations. Chains were run to ensure that the 45% lead to the maximum number of families with 1:1 ortho- minimum effective sample size was >50 and maximum rela- logues (5,394; supplementary fig. S1, Supplementary Material tive difference <0.3 in at least one sample. Results are sum- online) while settings with 40% identity and 80% coverage marized in supplementary table S2, Supplementary Material lead to 5,326 families with 1:1 orthologues (supplementary online, and figure S2, Supplementary Material online, shows fig. S1, Supplementary Material online). Since using a higher the six chains for the three samples. In addition to the con- coverage had only a cost of 68 core families, the stricter cri- vergence of the two chains for each sample, our results reveal teria were applied for family clustering. Families with at least extremely consistent results between samples. Figure 1A Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 631 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE FIG.1.—Phylogeny, divergence estimates, and number of genes under positive selection in five related smut fungal species parasitizing different host plants. (A) Chronogram of the five fungal pathogens as estimated under a relaxed molecular clock. Boxes represent 95% posterior intervals, with corresponding values indicated below. (B) Pairwise sequence differences, for both the noncoding genome and the proteome (nonsynonymous differences). (C) Number of positively selected genes on each terminal branch (total number of genes and genes predicted to encode a secreted protein). shows estimates from one chain of the third data set, which positive selection for each gene family. This is typically shows a minimum effective sample size >300. achieved by measuring the ratio of nonsynonymous versus synonymous substitutions (d /d ratio) using models of codon N S sequence evolution (Yang 2006). In particular, nonhomoge- Detection of Positive Selection neous models of sequence evolution estimate the d /d ratio N S For gene families with at least three members, translated independently in different lineages, yet at the cost of potential sequences were employed to create maximum likelihood phy- overparametrization issues. In the manual of the PAML pack- logenetic trees using PhyML 3.0 (Guindon et al. 2010)with a age, the authors state that such models should only be used minimum parsimony starting tree and the LG amino acid sub- for hypothesis testing and advise against using them for scans stitution model with a four-classes gamma distribution of site- of positive selection. Dutheil et al. (2012) proposed a model specific substitution rate (Le and Gascuel 2008). The best tree selection approach (implemented in the TestNH package) topology obtained from nearest neighbor interchange (NNI) allowing to select for the best nonhomogeneous model sup- and subtree pruning recrafting (SPR) searches was kept ported by the data. They start by fitting the simplest (homo- (Guindon et al. 2010). BppML (Dutheil and Boussau 2008) geneous) model and sequentially add parameters to model was then used to re-estimate branch lengths from the codon variation of selective regime among lineages. Because the alignment using the YN98 substitution model (Nielsen and number of possible models is large even for small data sets, Yang 1998). We next aimed at inferring the occurrence of two heuristic approaches have been introduced: the “free” 632 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE heuristic permits unconnected branches from the tree to pathogenic development (Kamper et al. 2006; Schirawski evolve under the same regime, whereas the “joint” heuristic et al. 2010). In total, these clusters contain 163 genes, where restricts model sharing to connected branches (see 100 reside in clusters without virulence phenotype and 63 Dutheil et al. 2012 for details). The choice of models to test reside in clusters with virulence phenotype upon deletion. is guided by statistics on the patterns of substitutions on the Both types of clusters contain each 32 genes for which a phylogenetic tree, an approach named substitution mapping d /d ratio could be determined (the missing genes are part N S (Romiguier et al. 2012). Apart from the model selection ap- of families that do not have at least three members and were proach, the underlying models of codon sequence evolution therefore not analyzed). The d /d ratios of all genes in clus- N S are identical to the one originally described by Yang ters were compared between clusters with and without viru- (Yang 1998; Yang and Nielsen 1998). Model selection was lence phenotype (Wilcoxon rank-sum test). performed with the TestNH software, which contains two programs: 1) MapNH (Romiguier et al. 2012) was used for Gene Ontology Terms Enrichment Analysis mapping substitutions on the previously inferred phylogenetic All proteins in S. reilianum f. sp. zeae,in S. reilianum f. sp. tree and 2) PartNH (Dutheil et al. 2012) was subsequently reilianum,and in U. hordei were considered for Gene employed to fit time nonhomogeneous models of codon sub- Ontology (GO) term enrichment analyses. GO terms were stitutions. PartNH uses the previously inferred substitution assigned using iprscan 1.1.0 (http://fgblab.org/runiprscan;de- maps in order to perform model comparisons and select a veloped by Michael R. Thon) which links GO information pro- nonhomogeneous model with minimal number of parame- vided by Interpro to each protein. In this way, 1,759 unique ters. Both methods “free” and “join” were compared in or- GO terms could be assigned to 4,130 proteins in S. reilianum der to scan for positive selection. Finally, putative secreted f. sp. zeae, 1,744 unique GO terms could be assigned to effector proteins were identified by predicting secretion using 4,124 proteins in S. reilianum f. sp. reilianum, and 1,757 SignalP 4.0 (Petersen et al. 2011) and proteins were consid- unique GO terms could be assigned to 3,922 proteins in U. ered as secreted if the program indicated the presence of a hordei (supplementary table S3, Supplementary Material on- signal peptide but no transmembrane domain. line). The Bioconductor package topGO (Alexaetal. 2006) To detect residues under positive selection in homologues wasthen usedtolink eachGO termtothe three major cat- of pit2, the branch-site model with Bayes Empirical Bayes egories “Cellular Component,” “Biological Process,” or (BEB) analysisasimplemented in PAML4(Yang 2007) was “Molecular Function.” Enrichment analysis was performed applied. We employed information about family composition, by computing P values for each GO term using Fisher’s classic alignment, and phylogeny as outlined above and defined test with parent–child correction (Grossmann et al. 2007). sr10529 and srs_10529 as foreground branches. A posterior Cytoplasmic proteins with and without signatures of positive probability threshold of> 95% was used for the BEB analysis. selection were compared for the three species separately, and differences were considered to be significant at the 5% level. Association of Positively Selected Genes with Repeats in U. hordei Strains and Growth Conditions We tested whether genes under positive selection are located The Escherichia coli derivative Top10 (Invitrogen, Karlsruhe, significantly closer to repetitive elements than average genes Germany) and the Saccharomyces cerevisiae strain BY4741 in thegenomeof U. hordei, which shows the highest content (MATa his3D1 leu2D met15D ura3D;Euroscarf,Frankfurt, of repetitive elements in the group of smut fungi investigated Germany; kindly provided by M. Bo ¨ lker, Marburg) were here. For this analysis, only a group of “uncharacterized in- used for cloning purposes. Sporisorium reilianum f. sp. zeae terspersed repeats” was investigated, because it was shown strains used in this study are listed in supplementary table S4, previously that this is the only category showing a strong as- Supplementary Material online. They are derivatives of the sociation with candidate effector genes (Dutheil et al. 2016). haploid solopathogenic strain JS161 which is capable of plant Binary logistic regressions were conducted in R using the rms colonization without the need of a mating partner, because it package (Harrell 2015). The “robcov” function of the rms expresses a compatible pheromone/receptor pair (Schirawski package was used in order to get robust estimates of each et al. 2010). Escherichia coli was grown in dYT liquid medium effect. The variable “distance to the closest interspersed (1.6% [w/v] Trypton, 1.0% [w/v] Yeast Extract [Difco (R)], repeat” was transformed by log(xþ 1) because of its extreme 0.5% [w/v] NaCl), or YT solid medium (0.8% [w/v] Trypton, distribution. 0.5% [w/v] Yeast Extract, 0.5% [w/v] NaCl, 1.3% [w/v] agar) supplemented with 100 mg/mL Ampicillin when needed. The Comparing d /d Ratios of Genes Residing in Virulence N S yeast S. cerevisiae was maintained in YPD solid medium (1% Clusters [w/v] yeast extract, 2% [w/v] (R)-Pepton, 2% [w/v] (R)-Agar, Previous work has identified several virulence gene clusters in 2% [w/v] glucose) and grown on SC URA medium (1.7% U. maydis and some of them play important roles during [w/v] Yeast Nitrogen Base without ammonium sulfate, Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 633 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE 0.147% [w/v] dropout-mix without Uracil, 2% [w/v] glucose) The drag and drop cloning method in yeast (Jansen et al. for selecting transformants containing the plasmid pRS426 2005) was used to generate plasmids pRS426 Dsr12968 Hyg (Sikorski and Hieter 1989) (kindly provided by M. Bo¨lker, #1, pRS426 Dsr14944 Hyg #2, pRS426 Dsr10059 Hyg #1, Marburg) or derivatives of pRS426. Strains of S. reilianum pRS426 Dsr10182 Hyg #1, pRS426 Dsr14558 Hyg #1, and were grown in liquid YEPS medium (1.0% [w/v] yeast ex- pRS426 Dsr12897 Hyg #1 which contain deletion constructs light tract, 0.4% [w/v] peptone, 0.4% [w/v] sucrose) at 28 Con a for deleting the candidate genes sr12968, sr14944, sr10059, rotary shaker at 200 rpm. sr10182, sr14558,or sr12897. These plasmids are a derivate of plasmid pRS426, which can be maintained in E. coli and S. cerevisiae (Sikorski and Hieter 1989). PCR-amplified left and right borders of each candidate gene and the hygromycin Construction of S. reilianum Strains resistance cassette were integrated in pRS426 by homologous Polymerase chain reactions were performed using the Phusion recombination in S. cerevisiae. Subsequently, the hygromycin High-Fidelity DNA Polymerase (New England Biolabs). resistance cassette was replaced with the Geneticin resistance Templates were either JS161 genomic or indicated plasmid cassette by ligation via SfiI restriction sites, yielding plasmids DNA. Restriction enzymes were obtained from New England pRS426 Dsr12968 Gen #1, pRS426 Dsr14944 Gen #3, Biolabs. Protoplast-mediated transformation was used to pRS426 Dsr10059 Gen #1, pRS426 Dsr10182 Gen #1, transform S. reilianum following a method established for pRS426 Dsr14558 Gen #1, and pRS426 Dsr12897 Gen #5, U. maydis (Schulz et al. 1990). Transformants were selected respectively. Gene deletion constructs were PCR-amplified on RegAgar plates (1.0% [w/v] yeast extract, 0.4% [w/v] from the respective plasmid using listed primers (supplemen- Bacto(R)-Pepton, 0.4% [w/v] Sucrose, 1 M Sorbitol, 1.5% tary table S5, Supplementary Material online). The obtained [w/v] Bacto(R)-Agar) supplemented with 200 mg/mL deletion constructs were used to transform the S. reilianum Geneticin and true resistance was tested by growing single strain JS161 to generate the gene deletion strains colonies on PD plates (3.9% [w/v] Potato-Dextrose Agar, 1% JS161Dsr12968, JS161Dsr14944, JS161Dsr10059, [v/v] Tris–HCl [1 M, pH 8.0]) supplemented with 50 mg/mL JS161Dsr10182, JS161Dsr14558, and JS161Dsr12897, Geneticin. Gene replacements with resistance markers were respectively. generated with a PCR-based method employing the previ- The drag and drop cloning method was also used to gen- ously described SfiI insertion cassette system (Brachmann erate plasmid pRS426 Dsr12084 Gen #1. PCR-amplified left et al. 2004; Kamper 2004) and were confirmed by and right borders of sr12084 and the Geniticin resistance Southern blot analysis. Genomic regions residing 1-kb up- cassette were integrated in pRS426 by homologous recombi- stream (left border) or downstream (right border) adjacent to nation in S. cerevisiae. The gene deletion construct for delet- open reading frames of candidate genes were PCR-amplified ing the candidate gene sr12084 was PCR-amplified from using the listed primer pairs (supplementary table S5, plasmid pRS426 Dsr12084 Gen #1 using primers Supplementary Material online) and genomic DNA of JS161 sr12084_lb_fw/sr12084_rb_rv and transformed into the S. as template. The resulting fragments were used for cloning reilianum strain JS161 to generate the gene deletion strain plasmids containing the respective deletion constructs. JS161Dsr12084. To obtain deletion constructs for the genes sr10529 and sr14347, PCR fragments containing the left and right borders Virulence Assays of each gene were ligated to the hygromycin resistance cas- sette of pBS-hhn (Kamper 2004) via SfiI restriction sites and The solopathogenic strain JS161 and deletion mutants cloned into pCRII-TOPO (Life Technologies) to generate thereof were grown in YEPS liquid medium to an op- light pTOPO Dsr10529 #1 and pTOPO Dsr14347 #1, respectively. tical density at 600 nm (OD ) of 0.8–1.0 and cell cultures Since the use of Geneticin as selection marker resulted in were adjusted to an OD of 1.0 with sterile water prior much less false positive transformants compared with the to injection into 1-week-old maize (Zea mays) seedlings of use of Hygromycin B, the hygromycin resistance cassettes in the dwarf cultivar “Gaspe Flint” (kindly provided by B. these plasmids were replaced by the Geneticin resistance cas- Burr, Brookhaven National Laboratories and maintained sette of pUMA 1057 (Brachmann et al. 2004) by ligation via by self-pollination). Plants were sowed in T-type soil of SfiI restriction sites, yielding plasmids pTOPO Dsr10529 G418 “Fruhstorfer Pikiererde” (HAWITA, Vechta, Germany) and pTOPO Dsr14347 Gen #1, respectively. Deletion con- and grown in a temperature-controlled greenhouse structs were PCR-amplified from plasmids pTOPO Dsr10529 (14 h-/10 h-light/dark cycle, with 28/20 C and 25,000– G418 and pTOPO Dsr14347 Gen #1 using the listed primers 90,000 lux during the light period). Virulence symptoms (supplementary table S5, Supplementary Material online) and were scored 9- to 10-week postinfection according to used to transform the S. reilianum strain JS161 to generate previously described symptoms (Ghareeb et al. 2011) the gene deletion strains JS161Dsr10529 and and the following categories were distinguished: the plant JS161Dsr14347, respectively. did not develop ears, the plant developed healthy ears 634 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE shorter or equal to 1 cm or the plant developed healthy (supplementary table S6, Supplementary Material online). ears longer than 1 cm, the plant developed spiky ears, Gene duplications in U. hordei have been hypothesized to be phyllody in ears or phyllody in tassels. Spore formation driven by mobile elements (Laurie et al. 2012). was only observed occasionally, and rarely the plant died due to the infection. Three independent infections were The Genomes of S. reilianum f. sp. zeae and S. reilianum f. carried out per strain, and at least three independent de- sp. reilianum Diverged around 1 Ma letion strains were tested for virulence. To establish a frame for our comparative analysis, we first calculated sequence similarity of the five smut fungi for non- coding intergenic and protein sequences. In addition, we es- Results timated divergence times by performing a molecular dating Candidate Effector Genes Are Less Conserved between analysis based on the core orthologues set of the five patho- Species Compared with Other Genes gens, using advanced models of protein sequence evolution We reconstructed families of homologous genes for the five and Bayesian inference as implemented in the PhyloBayes smut fungi U. hordei, U. maydis, S. scitamineum, S. reilianum f. package (Lartillot et al. 2009). As calibration point, we used sp. zeae,and S. reilianum f. sp. reilianum using the SiLiX clus- thedivergencetimeof U. hordei and U. maydis, previously tering algorithm (Miele et al. 2011). We optimized the cluster- estimated to be between 27 and 21 Myr (Bakkeren and ing parameters to maximize the occurrence of orthologues and Kronstad 2007). In alignable intergenic regions U. hordei minimize the number of paralogues within each family. In this shares 57% identity with S. reilianum f. sp. reilianum and way, we were able to reconstruct 8,761 families, among which 77% identity in protein sequences (fig. 1B). Monte-Carlo 5,266 had at least one gene in each species (supplementary Markov chains were run for three independent gene samples table S6, Supplementary Material online). As a consequence, totaling >20,000 amino acid positions each, and two chains we found at least one homologous sequence in four species for were run in each case to assess convergence. The resulting 78% of all genes. About 5,254 gene families are found to have posterior distribution of divergence times were used to infer exactly one member in each species and were therefore taken 95% posterior intervals. The split between U. maydis and the as true orthologues (referred to as “core orthologous set” in Sporisorium species was estimated to have occurred 20 Myr the following). Considering that secreted proteins are putative ago (95% posterior interval 25–12 Myr; fig 1A and supple- effectors, we used SignalP (Petersen et al. 2011)to predict mentary table S2, Supplementary Material online). secretion of the encoded protein for each gene (supplementary Sporisorium reilianum f. sp. reilianum shares 61% nucleotide table S1, Supplementary Material online). We report that 920 identity in alignable intergenic regions with U. maydis,and (11%) families contained only genes encoding a predicted se- 79% sequence identity at the protein level (fig. 1B). The di- creted protein, whereas 7,657 (87%) contained only genes vergence times of S. scitamineum andthe twoformae spe- encoding a protein not predicted to be secreted. The remaining ciales of S. reilianum were calculated to be 13 Myr ago (95% 184 (2%) families contained both predicted secreted and cy- posterior interval 19–7 Myr; fig. 1A and supplementary table toplasmic proteins (supplementary table S6, Supplementary S2, Supplementary Material online), which is consistent with Material online). The occurrence of families with both secreted the mean divergence estimated between the hosts sorghum and cytoplasmic proteins can be explained by 1) false negative and sugarcane (10 Myr with a posterior interval of 8–13 Myr, predictions for secretion, as truncated C-terminal sequences average over eight studies, source: timetree.org; Kumaretal. were not removed from the data set, 2) wrong gene annota- 2017). Sporisorium reilianum f. sp. reilianum and S. scitami- tions, or 3) gain or loss of a secretion signal peptide during neum share 74% noncoding nucleotide identity and 88% effector evolution (Poppe et al. 2015). Among all predicted identity at the protein level (fig. 1B). Finally, the two S. reilia- secreted proteins, 52% have at least one orthologue in all other num strains diverged 1.1 Myr ago (95% posterior interval 2.4– species, which is significantly less than the global 78% propor- 0.4 Myr; fig. 1A) and share 98% noncoding nucleotide iden- 2 16 tion for all proteins (v test, P value< 2.210 ). Genes tity and 99% protein identity (fig. 1B). We note that the encoding putative effector proteins are therefore less conserved estimation of this divergence date varied with the gene set across species than other genes, either because their sequence used, and was in some cases found to be older (1.7 Myr, with is evolving faster, preventing the recovery of homologous rela- a 95% posterior interval of 4.7–0.6 Myr, see supplementary tionships, or because effector genes are created or lost at a table S2, Supplementary Material online). The comparison of higher rate. In U. hordei, we observe several species-specific the five smut genomes therefore encompasses a broad evolu- family expansions. There were 17 families which encompassed tionary time, and the divergence times obtained are compatible 5–25 members, but no orthologue in other species (supple- with previous estimates from smaller data sets (Munkacsi et al. mentary table S6, Supplementary Material online). Moreover, 2007). The speciation times of the investigated smut species we identified three families with up to 62 members in U. hor- largely predate the 10,000 years of crop plant domestication, dei, but only one member in up to three of the other species which implies that adaptation to the agricultural host, if any, Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 635 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE will be negligible when interpreting the interspecific patterns of of earlier studies in other pathosystems that showed that sequence divergence, as it represents a marginal proportion of predicted secreted proteins are often under positive selection, thetimesince thedivergencefrom the ancestral species. which can be attributed to their direct interaction with host proteins (Joly et al. 2010; Wicker et al. 2013; Poppe et al. 2015). Notably, all genes found under positive selection in Sporisorium reilianum Contains the Largest Number of the two strains of S. reilianum,in S. scitamineum and in U. Positively Selected Genes maydis share orthologous genes in the other species (supple- To detect positive selection, 6,205 families with at least mentary tables S3 and S6, Supplementary Material online). In three members (orthologues and/or paralogues, see supple- contrast, genes with signs of positive selection in U. hordei mentary table S6, Supplementary Material online) were, re- belong largely (36 out of 49 genes) to families showing gardless of their species composition, aligned on the codon species-specific expansions (supplementary tables S3 and and amino acid level and a phylogenetic tree was inferred. S6, Supplementary Material online). Obtaining accurate alignments is critical for detecting positive selection since alignment errors frequently inflate the false discovery rate (Schneider et al. 2009; Jordan and Goldman Genes Under Positive Selection in U. hordei Are Associated 2012). We therefore developed a stringent bioinformatics with Uncharacterized Interspersed Repeats pipeline for the filtering of sequence alignments by masking ambiguous alignment positions for further analysis (see Among the species compared here, the genome of U. hordei Materials and Methods). To scan for positive selection, we shows the highest fraction of repetitive elements (Laurie et al. employed a nonhomogeneous model of sequence evolution 2012; Dutheil et al. 2016). Such elements are known to con- allowing d /d ratios to vary along the phylogeny, in combi- tribute to gene family expansions (Kazazian 2004), and have N S nation with two heuristic model selection procedures to avoid been suggested to contribute to adaptation by providing ad- overparametrization issues (Nielsen and Yang 1998; Dutheil vantageous mutations, for instance by repeat-induced point et al. 2012). Model parameters could not be fitted by either mutations (RIP) leakage which was revealed in a species com- one of the two methods in 1.7% of branches. The two model plex of Leptosphaeria (Rouxel et al. 2011; Grandaubert et al. selection procedures led to highly consistent estimates of 2014). As sequence signatures of RIP were found in LTR ele- branch-specific d /d ratios (Spearman’s rank correlation co- ments of U. hordei (Laurie et al. 2012), we tested whether N S efficient equal to 0.85, P value< 2.210 ). The distribution genes under positive selection in U. hordei are physically as- of d /d was highly skewed with a median value of 0.06, sociated with repetitive elements. We performed a binary lo- N S demonstrating the strong predominance of purifying selec- gistic regression with the prediction of positive selection as a tion throughout lineages and genes. The mean value of d / response variable (that is, whether the underlying branch has d ratios for lineages undergoing positive selection (d /d >1) a d /d ratio >1) and we considered three putative explana- S N S N S was 4.1 (median 1.9). Although a d /d ratio >1 is indicative tory variables for each analyzed gene: 1) whether the gene is N S of positive selection, the absolute value of the ratio is a poor predicted to encode a secreted protein, 2) whether the gene is indicator of the strength of undergoing selection. In particular, duplicated, and 3) the distance of the gene to the closest high ratio values can be obtained because of low d values, interspersed repeat. The complete linear model explains and the d rate might include neutral substitutions, such as 50% of the observed variance, and the three explanatory nonsynonymous substitution that are conservative regarding variables are all significant at the 0.1% level (supplementary certain biochemical properties of the amino acids involved table S7, Supplementary Material online). These results sug- (Sainudiin et al. 2005). The largest number of genes with signs gest that positively selected genes in U. hordei are associated of positive selection was found in S. reilianum f. sp. zeae (84 with duplication events, and positive selection is more likely to genes, of which 25 encode predicted secreted proteins) and occur at genes encoding putative effectors. In addition, S. reilianum f. sp. reilianum (111 genes of which 27 encode the proximity of interspersed repeats increases the odds of predicted secreted proteins) (fig. 1C). In addition, a substantial positive selection, independently of the two other effects, number of positively selected candidate genes was also found and is confirmed by a stratification approach: the effect in U. hordei (49, and of these, 22 genes are predicted to code still holds when only duplicated genes are considered, or for secreted proteins), but only very few in U. maydis (2 genes) only genes encoding a secreted protein, or the combina- and S. scitamineum (7 genes) (fig. 1C). A list of all proteins tion of the two (supplementary table S7, Supplementary with their associated d /d ratios in each species is provided in Material online). This finding corroborates previous results N S supplementary table S3, Supplementary Material online. obtained in other microbial plant pathogens where it was Predicted secreted proteins were significantly enriched in described that effector genes tend to localize in repeat the group of proteins under positive selection in U. hordei rich regions and where it was suggested that such regions and in the two investigated formae speciales of S. reilianum contribute to the rapid evolution of effector genes (P values< 10 ; Fisher’s exact test). This corroborates results (Raffaele and Kamoun 2012). 636 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Table 1 Gene Ontology Terms Significantly Overrepresented in Positively Selected Genes Encoding Cytoplasmic Proteins in Sporisorium reilianum and Ustilago hordei a b Gene Ontology ID Gene Ontology Description Category P values Species GO: 0030532 Small nuclear ribonucleoprotein complex CC 0.023 Srr GO: 0045263 Proton-transporting ATP synthase complex, coupling factor F(o) CC 0.038 and 0.036 Srr and Srz GO: 0044425 Membrane part CC 0.047 Srr GO: 0031668 Cellular response to extracellular stimulus BP 0.015 and 0.013 Srr and Srz GO: 0051186 Cofactor metabolic process BP 0.015 Srr GO: 0042594 Response to starvation BP 0.017 and 0.012 Srr and Srz GO: 0009267 Cellular response to starvation BP 0.020 and 0.014 Srr and Srz GO: 0006790 Sulfur compound metabolic process BP 0.022 Srr GO: 0051301 Cell division BP 0.023 Srr GO: 0006777 Mo-molybdopterin cofactor biosynthetic process BP 0.027 Srr GO: 0009605 Response to external stimulus BP 0.027 and 0.012 Srr and Srz GO: 0043545 Molybdopterin cofactor metabolic process BP 0.034 Srr GO: 0043413 Macromolecule glycosylation BP 0.036 Srr GO: 0022402 Cell cycle process BP 0.037 Srr GO: 0051189 Prosthetic group metabolic process BP 0.039 Srr GO: 0006139 Nucleobase-containing compound metabolic process BP 0.012 Srz GO: 0016070 RNA metabolic process BP 0.023 Srz GO: 1901360 Organic cyclic compound metabolic process BP 0.016 Srz GO: 0006725 Cellular aromatic compound metabolic process BP 0.024 Srz GO: 0046483 Heterocycle metabolic process BP 0.023 Srz GO: 0035383 Thioester metabolic process BP 0.041 Srr GO: 1902589 Single-organism organelle organization BP 0.048 Srr GO: 0015074 DNA integration BP 0.039 Uh GO: 0006259 DNA metabolic process BP 0.023 Uh GO: 0010181 FMN binding MF 0.047 and 0.022 Srr and Srz GO: 0030515 snoRNA binding MF 0.021 Srz GO: 0016491 Oxidoreductase activity MF 0.028 Srz GO: 0016853 Isomerase activity MF 0.029 Uh CC, cellular component; BP, biological process; MF, molecular function. Calculated by Fisher’s classic test with Parent–Child correction. Only entries with P value 0.05 are shown. Species in which GO terms were found enriched. Srr, S. reilianum f. sp. reilianum;Srz, S. reilinaum f. sp. zeae;Uh, U. hordei. Positively Selected Genes Encoding Cytoplasmic Proteins in analysis for cytoplasmic proteins under positive selection in U. S. reilianum and U. hordei hordei was conducted, but only led to top-level categories (DNA integration, DNA metabolic process, and isomerase ac- Although we expect effector genes to be under positive se- tivity; see table 1). lection, we find that the majority of positively selected genes in S. reilianum encodescytoplasmic proteins(fig. 1C). To as- sess the putative functional role of these genes, we performed Virulence Contribution of Effector Genes Showing Signs of a Gene Ontology term enrichment analysis, comparing cyto- Positive Selection in S. reilianum plasmic proteins under positive selection to cytoplasmic pro- Candidate effector genes inferred to be under positive selec- teins not under positive selection (table 1). This analysis tion in a particular species could play a critical role in patho- revealed that genes with a potential role in metabolic pro- genicity. Therefore, we sought to assess the contribution to cesses, like sulfur compound metabolism, molybdopterin co- virulence of such candidate genes by creating individual dele- factor metabolic process, RNA metabolic process, organic tion mutants. In total, we tested nine candidate genes with cyclic compound metabolic process, and oxidoreductase ac- high d /d ratios and predicted to encode secreted proteins: N S tivity, as well as responses to starvation and extracellular stim- three with signatures of positive selection only in S. reilianum uli are significantly overrepresented at the 5% level (Fisher’s f. sp. zeae, three with signatures of positive selection in S. classic test with Parent–Child correction; see table 1). This reilianum f. sp. zeae as well as in S. reilianum f. sp. reilianum could indicate that cytoplasmic proteins under positive selec- and three with signatures of positive selection only in S. rei- tion contribute to metabolic changes which might be needed lianum f. sp. reilianum. All nine chosen candidate genes to- to survive with the limited nutrients available on the surface or gether with their characteristics are summarized in table 2. in the biotrophic interface of different host plants. A similar Deletion mutants were generated in the haploid Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 637 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE Table 2 Positively Selected Genes That Were Deleted in Sporisorium reilianum f. sp. zeae and Their Selection Criteria Family Gene ID Gene Description d /d Species with Number of Virulence Phenotype Closest Ortholog in N S Detected Positive Paralogs of Candidate Ustilago maydis a b Selection in Srz Gene Deletion FAM001428 sr10529 Conserved hypo- 31.1469 Srz and Srr 0 Virulence abolished UMAG_01375 (pit2) thetical protein FAM005472 sr10059 Conserved hypo- 6.53881 Srz and Srr 0 Virulence unaffected UMAG_05306 (cluster 19A) thetical Ustilaginaceae- specific protein FAM000532 sr10182 Conserved hypo- 1.57473 Srr 10 Virulence unaffected UMAG_00492 thetical protein FAM002067 sr12968 Conserved hypo- 37.9007 Srr 0 Virulence unaffected UMAG_02006 thetical protein FAM003728 sr14558 Conserved hypo- 24.355 Srz 0 Virulence unaffected UMAG_03564 thetical protein FAM004113 sr14944 Conserved hypo- 4.30527 Srz and Srr 0 Virulence unaffected UMAG_04034 (cluster 11–16) thetical Ustilaginaceae- specific protein FAM003465 sr14347 Conserved hypo- 544.37 Srz 5 Virulence unaffected UMAG_03349 thetical protein FAM001868 sr12897 Conserved hypo- 999 Srr 0 Virulence unaffected UMAG_01820 thetical protein e h k FAM000842 sr12084 Conserved hypo- 999 Srz 2 Virulence unaffected UMAG_00792 (cluster 1–32) thetical protein Species are S. reilianum f. sp. zeae (Srz) and S. reilianum f. sp. reilianum (Srr). Based on BLASTp search with an e-value cutoff of 0.001. BasedonBLASTpsearch. Maximum value allowed by the PartNH program. Infinity due to low value of d . The ten paralogs include: sr13431, sr11876, sr16607, sr11405, sr10621, sr16723, sr16877, 13293, sr11163.2,and sr15970. The five paralogs include: sr12257, sr11661, sr13976, sr14607,and sr11273. The two paralogs include: sr12085 and sr12086. As described in Doehlemann et al. (2011) and Mueller et al. (2013). As described in Ka €mperetal. (2006). As described in Schirawski et al. (2010). solopathogenic strain JS161 of S. reilianum f. sp. zeae. The gene sr10529 in S. reilianum f. sp. zeae is orthologous This strain is capable of colonizing maize plants and cause to the previously identified and characterized gene pit2 disease without a compatible mating partner (Schirawski (UMAG_01375)in U. maydis. et al. 2010) but virulence is much reduced relative to in- Pit2 plays an essential role in virulence as inhibitor of a fection with mating-compatible wild-type strains and group of maize papain-like cysteine proteases that are spores are only rarely produced. Deletion mutants were secreted to the apoplast (Doehlemann et al. 2011; also generated in strain JS161 in cases where positive se- Muelleretal. 2013). Previous work identified a conserved lection was only detected in S. reilianum f. sp. reilianum domain of 14 amino acids (PID14) in Pit2 as required and (table 2), because no solopathogenic strain is presently sufficient for the inhibition of maize cysteine proteases available for S. reilianum f. sp. reilianum. For each gene, (Muelleretal. 2013). When the branch-site model of at least three independent deletion mutants were gener- PAML 4 (Yang 2007) wasusedtoidentify amino acid ated and tested for virulence. To determine virulence, residues under positive selection in the Pit2 orthologues Gaspe Flint, a dwarf variety of corn, was infected and of the two S. reilianum species, only two residues residing symptoms were scored in male and female flowers in the PID14 domain were found under positive selection. (fig. 2). Only the deletion of sr10529, a gene showing However, 24 positively selected residues were detected positive selectioninbothformae speciales of S. reilianum, outside this domain in the 57 amino acid long C-terminal showed a strong reduction in virulence (table 2 and fig. 2). part (fig. 3). 638 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE n = 83 n = 88 n = 29 n = 29 JS161 #G3 #G4 #G5 123 4 JS161ΔSr10529 (Pit2) B F n = 95 n = 100 n = 95 n = 87 n = 93 n = 96 n = 95 n = 102 40 40 0 0 JS161 #3 #4 #5 JS161 #2 #3 #4 123 4 123 4 JS161ΔSr12968 JS161ΔSr14558 C G n = 95 n = 103 n = 103 n = 90 n = 83 n = 89 n = 85 n = 89 100 100 80 80 60 60 40 40 20 20 0 0 JS161 #15 #28 #31 JS161 #1 #2 #3 12 3 4 12 34 JS161ΔSr14944 JS161ΔSr14347 D H n = 95 n = 100 n = 95 n = 102 n = 95 n = 103 n = 103 n = 90 100 100 80 80 60 60 40 40 20 20 0 0 JS161 #25 #40 #47 JS161 #1 #2 #3 123 4 123 4 JS161ΔSr10059 JS161ΔSr12897 E I n = 92 n = 94 n = 93 n = 91 n = 94 n = 77 n = 94 n = 87 n = 82 20 20 0 0 JS161 #1 #2 #5 JS161 #4 #6 #9 #11 1234 12 3 4 5 JS161ΔSr10182 JS161ΔSr12084 plant without ears healthy ears ≤ 1 cm healthy ears > 1 cm spiky ears phyllody in ears ≤ 1 cm phyllody in ears > 1 cm phyllody in tassels spore formation dead plants FIG.2.—Virulence phenotypes of single deletion mutants of positively selected genes in Sporisorium reilianum f. sp. zeae. Plants of the maize variety “Gaspe Flint” were infected with the solopathogenic strain JS161 or independent deletion mutants of candidate genes as indicated below each bar. Deletion Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 639 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Schweizer et al. GBE FIG.3.—Distribution of positively selected amino acids in the cysteine protease inhibitor Pit2. The alignment shows the protein sequences of orthologues in Ustilago hordei (UHOR_02064), U. maydis (UMAG_01375), Sporisorium scitamineum (SPSC_03677), S. reilianum f. sp. zeae (sr10529), and S. reilianum f. sp. reilianum (srs_10529). Sites under positive selection detected by a branch-site model are indicated by colored bold letters. Residues colored in red indicate positive selection detected in the respective species and purple residues indicate sites found under positive selection in both species. The yellow shaded area is orthologous to the previously identified conserved PID14 domain, which is required and sufficient for inhibition of a group of papain-like cysteine proteases. Green sequences indicate secretion signal peptides and bold numbers above the alignment indicate positions in UHOR_02064. positive selection over time spans of several millions of years Discussion is notoriously difficult (Gillespie 1994) because of two main We used evolutionary comparative genomics of five related reasons: 1) periods where genes are evolving under positive smut fungi infecting four different host plants to identify selection may occur episodically and may be followed by long genes with a signature of positive selection during species episodes of purifying selection, leading to an average d /d N S divergence, with a focus on genes encoding predicted se- <1 on long periods of time and 2) fast evolving genes may creted proteins, as such genes were suggested to contribute diverge to an extent where their homology is difficult to infer to virulence in various plant pathogenic microbes (Aguileta and where they can no longer be aligned reliably. To over- et al. 2010; Stukenbrock et al. 2011; Hacquard et al. 2012; come this problem, more genome information of species with Dong et al. 2014; Huang et al. 2014; Sharma et al. 2014).Our intermediate branching points is needed (Gillespie 1994). analysis revealed that positive selection is found between Predicted secreted proteins were about three times over- paralogous genes in U. hordei, where they belong to families represented in the set of positively selected genes, which with species-specific expansions. In contrast, genes under illustrates the importance of secreted proteins in adaptation positive selection in the other four species belong to families processes of smut fungi. This also corroborates results in other of orthologous sequences. Although we find evidence for a plant pathogenic microbes like Melampsora sp., Z. tritici and large set of genes under positive selection in the S. reilianum the wheat powdery mildew Blumeria graminis (Joly et al. species, signatures for positive selection are hardly detectable 2010; Stukenbrock et al. 2011; Wicker et al. 2013). in the more distant relatives U. hordei, U. maydis,and S. However, the majority of positively selected genes encodes scitamineum that diverged earlier. Finding evidence for FIG.2. Continued of sr10529 led toa strong reductioninvirulence (A). In contrast, deletion of the candidate genes sr12968 (B), sr14944 (C), sr10059 (D), sr10182 (E), sr14558 (F), sr14347 (G), sr12897 (H), and sr12084 (I) did not alter virulence. Symptoms were scored 9-week postinfection and categorized according to severeness as illustrated in the legend below the bar plots. Results are shown as mean of three independent experiments in relation to the total number of infected plants, which is indicated above each bar (n). Note that strains JS161DSr10529 #G4 and #G5 (A) were only infected in one replicate. 640 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE cytoplasmic proteins (fig. 1C), suggesting that both secreted due to the weak infection behavior of JS161. For example, and nonsecreted proteins are important targets of adaptation. spore formation is only rarely observed after infecting maize A Gene Ontology analysis in S. reilianum showed that mainly plants with the solopathogenic strain. In contrast, infections processes related to metabolism and its regulation as well as resulting from infections with two compatible haploid strains responses to starvation and external stimuli are enriched in show spores in 40% of the infected plants (Zuther et al. cytoplasmic proteins under positive selection. This points at a 2012). This means that defects related to spore formation will role of these proteins in adaptation to differences in nutrient not be evident in mutants of JS161. In three cases, positive availability in the respective host plants maize and sorghum as selection was detected in orthologous genes in S. reilianum f. well as responses to cues originating from the respective host sp. reilianum while candidate effector genes were for exper- (Haueisen and Stukenbrock 2016). A study conducted in U. imental reasons deleted in S. reilianum f. sp. zeae. Therefore, maydis has shown that the fungus induces major metabolic it cannot be excluded that these effectors might have a viru- changes in the host plant upon infection during establishment lence function in S. reilianum f. sp. reilianum. In this case, the of biotrophy and undergoes a series of developmental tran- positively selected effector genes might have evolved during sitions during host colonization that are likely influenced by adaptation to the sorghum host and present host specificity the host environment (Doehlemann et al. 2008). It is thus genes. In summary, our virulence assays leave open the pos- conceivable that the two S. reilianum accessions have adapted sibility that the eight candidate genes which did not show a to their different hosts that differ significantly for example in contribution to virulence could play a role in pathogenicity their amino acid and vitamin composition (Etuk et al. 2012). under conditions not tested here. Alternatively, candidate ef- Furthermore, recent studies in U. maydis suggested that in- fector proteins might also be positively selected for traits that tracellular changes of metabolism influence virulence, and are not directly linked to pathogenicity. Such traits could for therefore the underlying proteins could be targets of positive instance involve competition with large numbers of other selection (Kretschmer et al. 2012; Goulet and Saville 2017). plant colonizing microbes (Zhan and McDonald 2013; Out of nine deletions of positively selected genes, only one Rovenich et al. 2014). Secreted proteins of S. reilianum could mutant, lacking sr10529, was affected in virulence. Although act for example as toxin or could efficiently utilize resources six of the deleted genes are single genes in S. reilianum f. sp. from the environment and thereby limit the growth of other zeae for which we failed to identify paralogs, sr12084 has two microbes. In these cases, a contribution to virulence is not paralogs, sr14347 has five paralogs, and sr10182 has ten expected to be observed in the employed infection assay paralogs. We restricted our analyses to generating deletion with the effector gene mutants. Moreover, our molecular dat- mutants in some of the genes under positive selection. This ing analysis showed that the common ancestors of the inves- leaves open the possibility that the paralogous genes have tigated smut species originated before the beginning of crop redundant functions in virulence. Adapting the CRISPR-Cas9 domestication. Therefore, positive selection, whose signs we technology allowing multiplexing (Schuster et al. 2017)to S. detect by our approach, has most likely occurred on ancestral reilianum will be instrumental in testing this hypothesis in fu- host plants and not on the domesticated host maize. ture studies. Alternatively, the candidate effectors we investi- Consequently, some of the candidate effector genes under gated may be needed under conditions which differ from positive selection might not be important for the colonization those tested here. For example, S. reilianum f. sp. zeae can of crop plants, but for infection of related wild species. also systemically colonize maize plants via root infection In U. maydis, we note that effector genes residing in clus- (Mazaheri-Naeini et al. 2015), a colonization route we have ters whose deletion affected virulence (Kamper et al. 2006; not assessed in our experimental setup. Moreover, we Schirawski et al. 2010) have similar d /d ratios as effector N S employed only one maize cultivar for infection assays. genes in clusters where the deletion had no effect on viru- Results from other pathosystems suggest that virulence lence (median d /d ratio 0.0619 vs. 0.1094; Wilcoxon rank N S effects can strongly depend on the host and pathogen gen- test with P value¼ 0.1848). Furthermore, orthologues of otypes, in particular in the presence of R and avr genes (Petit- the effectors Pep1, Stp1, and Cmu1, which were shown Houdenot and Fudal 2017). No avr-R gene interaction was to have important roles in pathogenicity of U. maydis described so far in the S. reilianum f. sp. zeae-maize pathos- (Doehlemann et al. 2009; Schipper 2009; Djamei et al. ystem. Instead, quantitative virulence differences are observed 2011) showed no signature of positive selection. These when different host cultivars are infected (Lu ¨ bberstedt et al. observations could suggest that certain fungal effector 1999). Knowing the expression profile of effector genes may proteins are under evolutionary constraint and are there- assist the identification of differences in development of the fore not free to accumulate nonsynonymous mutations. mutants compared with wild-type strains. Since we lack this Such effectors are conserved over long time spans information, we scored disease symptoms only in the inflor- (Schirawski et al. 2010; Hemetsberger et al. 2015; escences 9 weeks after infection. Additionally, it may be Sharma et al. 2015) and this illustrates that they are in- possible that small differences in virulence between JS161 strumental for successful infections in a large group of and deletion mutants of candidate genes remain undetected smut fungi. They probably target molecules shared by Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 641 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE several host plants, for example, housekeeping functions of positively selected genes did not encode predicted secreted that cannot easily evolve in response to the binding of an proteins. Our results leave open the possibility that many genes effector. with signatures of positive selection contribute to virulence un- One candidate gene (sr10529) under positive selection in der conditions not tested in this study or are selected in traits both formae speciales of S. reilianum showed a strong con- that are not directly related to pathogenicity. tribution to virulence upon deletion. It is orthologous to the previously described protease inhibitor Pit2 in U. maydis, Supplementary Material where the deletion also abolished virulence (Doehlemann Supplementary data are available at Genome Biology and et al. 2011; Mueller et al. 2013). Positively selected residues Evolution online. in the PID14 domain of Pit2 might reflect that different pro- teases need to be inhibited in maize and sorghum. Pit2 might thus contribute to determining the host range of the respec- tive species. A role of cysteine protease inhibitors in host spe- cificity was demonstrated in Phytophthora infestans,a Acknowledgments pathogen of potato and its sister species P. mirabilis,which Our work was supported through the LOEWE program of the infects the ornamental plant Mirabilis jalapa. Positively se- state of Hesse through SYNMICRO and through the Max lected orthologous protease inhibitors were shown to inhibit Planck Society. G.S. was funded by the International Max proteases specific to the respective host plants and this spe- Planck Research School (IMPRS) for Molecular, Cellular and cificity could be traced back to a single amino acid substitution Environmental Microbiology. We thank Michael Bo¨lker for (Dong et al. 2014). Surprisingly, 24 positively selected sites in providing the yeast strain BY4741 and the plasmid pRS426 Pit2 were detected outside the PID14 domain in the 57 amino which were used for cloning purposes, and Benoit Nabholz acid long C-terminal part in both S. reilianum f. sp. zeae and S. and Nicolas Lartillot for guidance with the dating analysis. We reilianum f. sp. reilianum. This finding raises the intriguing are grateful to Markus Rampp (Max Planck Computing possibility that the C-terminus of Pit2 might possess a second Center Garching) for assistance with computational work function that is independent of protease inhibition. Earlier and to Stefan Schmidt for expert greenhouse service, as work has shown that the pit1 gene encoding a transmem- well as all present and former group members for lively dis- brane protein is located next to the pit2 effector gene and cussions and Eva H. Stukenbrock for her comments on the both genes contribute similarly to virulence (Doehlemann manuscript. et al. 2011). Furthermore, pit1 and pit2 are divergently tran- scribed, which makes it likely that the expression of pit1 and pit2 is coregulated. In addition, this gene arrangement of pit1 Literature Cited and pit2 is conserved in U. hordei, U. maydis, S. scitamineum, Aguileta G. 2012. Genes under positive selection in a model plant path- and S. reilianum (Sharma et al. 2015). This finding has led to ogenic fungus, Botrytis. Infect Genet Evol. 12(5):987–996. the speculation that Pit1 and Pit2 somehow act together to Aguileta G, et al. 2010. Finding candidate genes under positive selection in non-model species: examples of genes involved in host specialization govern virulence of U. maydis and related smut fungi. It was in pathogens. Mol Ecol. 19(2):292–306. hypothesized that that Pit2 shuttles apoplastic maize proteins Aguileta G, Refregier G, Yockteng R, Fournier E, Giraud T. 2009. Rapidly toward Pit1, thereby scavenging damage-associated mole- evolving genes in pathogens: methods for detecting positive selection cules (Doehlemann et al. 2011). In this scenario, the positively and examples among fungi, bacteria, viruses and protists. Infect Genet selected amino acids in the C-terminus of Pit2 could have Evol. 9(4):656–670. Alexa A, Rahnenfu ¨ hrer J, Lengauer T. 2006. Improved scoring of functional been selected for scavenging such molecules as adaptation groups from gene expression data by decorrelating GO graph struc- to the two hosts. In future studies, it will be highly interesting ture. Bioinformatics 22(13):1600–1607. to complement the pit2 mutant of S. reilianum f. sp. zeae with Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local the pit2 orthologue of S. reilianum f. sp. reilianum to seeifthis alignment search tool. J Mol Biol. 215(3):403–410. promotes virulence on sorghum. Anikster Y. 1984. The Formae Speciales. In: Bushnell WR, Roelfs AP, edi- tors. The cereal rusts. London: Academic Press. Bagchi R, et al. 2014. Pathogens and insect herbivores drive rainforest plant diversity and composition. Nature 506(7486):85–88. Conclusions Bakkeren G, Kronstad JW. 2007. Bipolar and Tetrapolar Mating Systems in Screens for genes with signs of positive selection are commonly the Ustilaginales. In: Heitman J, Kronstad JW, Taylor J, Casselton L, used to identify candidate effector genes in various plant path- editors. Sex in fungi. Washington: ASM Press. Begerow D, et al. 2014. Ustilaginomycotina. In: Esser K, editor. The ogenic microbes. However, it is currently largely open whether Mycota – a comprehensive treatise on fungi as experimental systems positively selected effector genes play indeed a role in virulence. for basic and applied research. Systematics and evolution (Part A). Here, we used comparative genomics of five smut fungi and Heidelberg: Springer. showed that only one out of nine genes under positive selection Blanchette M, et al. 2004. Aligning multiple genomic sequences with the contributes to virulence of S. reilianum. Moreover, the majority Threaded Blockset Aligner. Genome Res. 14(4):708–715. 642 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Brachmann A, Ko ¨ nig J, Julius C, Feldbru ¨ gge M. 2004. A reverse genetic Grossmann S, Bauer S, Robinson PN, Vingron M. 2007. Improved detec- approach for generating gene replacement mutants in Ustilago may- tion of overrepresentation of Gene-Ontology annotations with parent dis. Mol Genet Genomics 272(2):216–226. child analysis. Bioinformatics 23(22):3024–3031. Brown JKM, Tellier A. 2011. Plant-parasite coevolution: bridging the Guindon S, et al. 2010. New algorithms and methods to estimate gap between genetics and ecology. Annu Rev Phytopathol. maximum-likelihood phylogenies: assessing the performance of 49:345–367. PhyML 3.0. Syst Biol. 59(3):307–321. Charif D, Lobry JR. 2007. SeqinR 1.0-2: a contributed package to the R Hacquard S, et al. 2012. A Comprehensive Analysis of Genes Encoding project for statistical computing devoted to biological sequences re- Small Secreted Proteins Identifies Candidate Effectors in Melampsora trieval and analysis. In: Bastolla U, Porto M, Roman HE, Vendruscolo larici-populina (Poplar Leaf Rust). Mol Plant Microbe in 25(3):279–293. M, editors. Structural approaches to sequence evolution: molecules, Harrell FE Jr. 2015. Regression modeling strategies. Switzerland: Springer networks, populations. New York: Springer Verlag. p. 207–232. Series in Statistics. de Jonge R, Bolton MD, Thomma BP. 2011. How filamentous pathogens Haueisen J, Stukenbrock EH. 2016. Life cycle specialization of filamentous co-opt plants: the ins and outs of fungal effectors. Curr Opin Plant Biol. pathogens – colonization and reproduction in plant tissues. Curr Opin 14(4):400–406. Microbiol. 32:31–37. Dean R, et al. 2012. The top 10 fungal pathogens in molecular plant Hemetsberger C, et al. 2015. The fungal core effector Pep1 is con- pathology. Mol Plant Pathol. 13(4):414–430. served across smuts of dicots and monocots. New Phytol. Djamei A, et al. 2011. Metabolic priming by a secreted fungal effector. 206(3):1116–1126. Nature 478(7369):395–398. Huang J, Si W,Deng Q,Li P,Yang S.2014. Rapid evolution of Doehlemann G, et al. 2008. Reprogramming a maize plant: transcriptional avirulence genes in rice blast fungus Magnaporthe oryzae. and metabolic changes induced by the fungal biotroph Ustilago may- BMC Genet. 15:45. dis. Plant J. 56(2):181–195. Jansen G, Wu C, Schade B, Thomas DY, Whiteway M. 2005. Drag&Drop Doehlemann G, et al. 2009. Pep1, a secreted effector protein of Ustilago cloning in yeast. Gene 344:43–51. maydis, is required for successful invasion of plant cells. PLoS Pathog. Joly DL, Feau N, Tanguay P, Hamelin RC. 2010. Comparative analysis of 5(2):e1000290. secreted protein evolution using expressed sequence tags from four Doehlemann G, Reissmann S, Aßmann D, Fleckenstein M, Kahmann R. poplar leaf rusts (Melampsora spp.). BMC Genomics 11:422. 2011. Two linked genes encoding a secreted effector and a mem- Jordan G, Goldman N. 2012. The effects of alignment error and alignment brane protein are essential for Ustilago maydis-induced tumour for- filtering on the sitewise detection of positive selection. Mol Biol Evol. mation. Mol Microbiol. 81(3):751–766. 29(4):1125–1139. Dong S, et al. 2014. Effector specialization in a lineage of the Irish potato Kamper J. 2004. A PCR-based system for highly efficient generation of famine pathogen. Science 343(6170):552–555. gene replacement mutants in Ustilago maydis. Mol Genet Genomics Dutheil JY, Boussau B. 2008. Non-homogeneous models of sequence 271(1):103–110. evolution in the Bioþþ suite of libraries and programs. BMC Evol K€ amper J, et al. 2006. Insights from the genome of the biotrophic fungal Biol. 8:255. plant pathogen Ustilago maydis. Nature 444(7115):97–101. Dutheil JY, et al. 2012. Efficient selection of branch-specific models of Kazazian HH. 2004. Mobile elements: drivers of genome evolution. sequence evolution. Mol Biol Evol. 29(7):1861–1874. Science 303(5664):1626–1632. Dutheil JY, et al. 2016. A tale of genome compartmentalization: the evo- Khrunyk Y, Mu ¨ nch K, Schipper K, Lupas AN, Kahmann R. 2010. The use of lution of virulence clusters in smut fungi. Genome Biol Evol. FLP-mediated recombination for the functional analysis of an effector 8(3):681–704. gene family in the biotrophic smut fungus Ustilago maydis.New Etuk EB, et al. 2012. Nutrient composition and feeding value of Sorghum Phytol. 187(4):957–968. for livestock and poultry: a review. J Anim Sci Adv. 2:510–524. Kretschmer M, Klose J, Kronstad JW. 2012. Defects in mitochondrial and Fisher MC, et al. 2012. Emerging fungal threats to animal, plant and peroxisomal b-oxidation influcence virulence in the maize pathogen ecosystem health. Nature 484(7393):186–194. Ustilago maydis. Eukaryot Cell 11(8):1055–1066. Franceschetti M, et al. 2017. Effectors of filamentous plant pathogens: Kumar S, Stecher G, Suleski M, Hedges SB. 2017. TimeTree: a resource for commonalities amid diversity. Microbiol Mol Biol R timelines, timetrees, and divergence times. Mol Biol Evol. 81:e00066–e00016. 34(7):1812–1819. Gehrig H, Schussler A, Kluge M. 1996. Geosiphon pyriforme, a fungus Lanver D, et al. 2017. Ustilago maydis effectors and their impact on viru- forming endocytobiosis with Nostoc (cyanobacteria), is an ancestral lence. Nat Rev Microbiol. 15(7):409–421. member of the Glomales: evidence by SSU rRNA analysis. J Mol Lartillot N, Lepage T, Blanquart S. 2009. PhyloBayes 3: a Bayesian software Evol. 43(1):71–81. package for phylogenetic reconstruction and molecular dating. Ghareeb HA, Becker A, Iven T, Feussner I, Schirawski J. 2011. Sporisorium Bioinformatics 25(17):2286–2288. reilianum infection changes inflorescence and branching architectures Laurie JD, et al. 2012. Genome comparison of barley and maize smut fungi of maize. Plant Physiol. 156(4):2037–2052. reveals targeted loss of RNA silencing components and species-specific Gillespie JH. 1994. The causes of molecular evolution. Oxford: Oxford presence of transposable elements. Plant Cell 24(5):1733–1745. University Press. Le SQ, Gascuel O. 2008. An improved general amino acid replacement Giraldo MC, Valent B. 2013. Filamentous plant pathogen effectors in ac- matrix. Mol Biol Evol. 25(7):1307–1320. tion. Nat Rev Microbiol. 11(11):800–814. Lo Presti L, et al. 2015. Fungal effectors and plant susceptibility. Annu Rev Goulet KM, Saville BJ. 2017. Carbon acquisition and metabolic changes Plant Biol. 66:513–545. during fungal biotrophic plant pathogenesis: insights from Ustilago Lo ¨ ytynoja A, Goldman N. 2008. Phylogeny-aware gap placement prevents maydis. Can J Plant Pathol. 39(3):247–266. errors in sequence alignment and evolutionary analysis. Science Grandaubert J, et al. 2014. Transposable element-assisted evolution and 320(5883):1632–1635. adaptation to host plant within the Leptosphaeria maculans- Lu ¨ bberstedt T, Xia XC, Tan G, Liu X, Melchinger AE. 1999. QTL mapping Leptosphaeria biglobosa species complex of fungal pathogens. BMC of resistance to Sporisorium reiliana in maize. Theor Appl Genet. 99(3– Genomics 15:891. 4):593–598. Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 643 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE Martin FM, Uroz S, Barker DG. 2017. Ancestral alliances: plant mu- histocompatibility complex and the SRK of the plant sporophytic tualistic symbioses with fungi and bacteria. Science self-incompatibility system. J Mol Evol. 60(3):315–326. 356(6340):eaad4501. Schipper K. 2009. Charakterisierung eines Ustilago maydis Genclusters, Martinez-Espinoza AD, Garcia-Pedrajas MD, Gold SE. 2002. The das fu ¨ r drei neuartige sekretierte Effektoren kodiert. Philipps- Ustilaginales as plant pests and model systems. Fungal Genet Biol. Universitat Marburg, https://d-nb.info/1000245756/34. 35(1):1–20. Schirawski J, et al. 2010. Pathogenicity determinants in smut fungi Mazaheri-Naeini M, Sabbagh SK, Martinez Y, S ejalon-Delmas N, Roux C. revealed by genome comparison. Science 330(6010):1546–1548. 2015. Assessment of Ustilago maydis as a fungal model for root in- Schneider A, et al. 2009. Estimates of positive Darwinian selection are fection studies. Fungal Biol. 119(2–3):145–153. inflated by errors in sequencing, annotation, and alignment. Miele V, Penel S, Duret L. 2011. Ultra-fast sequence clustering from sim- Genome Biol Evol. 1:114–118. ilarity networks with SiLiX. BMC Bioinformatics 12:116. Schulz B, et al. 1990. The b alleles of U. maydis, whose combinations Mo ¨ ller M, Stukenbrock EH. 2017. Evolution and genome architecture in program pathogenic development, code for polypeptides containing fungal plant pathogens. Nat Rev Microbiol. 15(12):756–771. a homeodomain-related motif. Cell 60:295–306. Mueller AN, Ziemann S, Treitschke S, Aßmann D, Doehlemann G. 2013. Schuster M, Schweizer G, Kahmann R. 2017. Comparative analyses of Compatibility in the Ustilago maydis–maize interaction requires inhibi- secreted proteins in plant pathogenic smut fungi and related basidio- tion of host cysteine proteases by the fungal effector Pit2. PLoS mycetes. Fungal Genet Biol. 16:1087–1845. Pathog. 9(2):e1003177. Schuster M, Schweizer G, Reissmann S, Kahmann R. 2016. Genome edit- Munkacsi AB, Stoxen S, May G. 2007. Domestication of maize, sorghum, ing in Ustilago maydis using the CRISPR–Cas system. Fungal Genet and sugarcane did not drive the divergence of their smut pathogens. Biol. 89:3–9. Evolution 61(2):388–403. Sharma R, Mishra B, Runge F, Thines M. 2014. Gene loss rather than gene Nielsen R. 2005. Molecular signatures of natural selection. Annu Rev gain is associated with a host jump from monocots to dicots in the Genet. 39:197–218. smut fungus Melanopsichium pennsylvanicum. Genome Biol Evol. Nielsen R, Yang Z. 1998. Likelihood models for detecting positively se- 6(8):2034–2049. lected amino acid sites and applications to the HIV-1 envelope gene. Sharma R, Xia X, Riess K, Bauer R, Thines M. 2015. Comparative genomics Genetics 148(3):929–936. including the early-diverging smut fungus Ceraceosorus bombacis Parniske M. 2008. Arbuscular mycorrhiza: the mother of plant root endo- reveals signatures of parallel evolution within plant and animal patho- symbioses. Nat Rev Microbiol. 6(10):763–775. gens of fungi and oomycetes. Genome Biol Evol. 7(9):2781–2798. Petersen TN, Brunak S, von Heijne G, Nielsen H. 2011. SignalP 4.0: dis- Sikorski RS, Hieter P. 1989. A system of shuttle vectors and yeast host criminating signal peptides from transmembrane regions. Nat strains designed for efficient manipulation of DNA in Saccharomyces Methods 8(10):785–786. cerevisiae. Genetics 122:19–27. Petit-Houdenot Y, Fudal I. 2017. Complex interactions between fungal Sperschneider J, et al. 2014. Diversifying selection in the wheat stem rust avirulence genes and their corresponding plant resistance genes and fungus acts predominantly on pathogen-associated gene families and consequences for disease resistance management. Front Plant Sci. reveals candidate effectors. Front Plant Sci. 5:372. 8:1072. Sperschneider J, et al. 2015. Genome-wide analysis in three Fusarium Plissonneau C, et al. 2017. Using population and comparative genomics to pathogens identifies rapidly evolving chromosomes and genes associ- understand the genetic basis of effector-driven fungal pathogen evo- ated with pathogenicity. Genome Biol Evol. 7(6):1613–1627. lution. Front Plant Sci. 8:119. Stergiopoulos I, de Wit PJ. 2009. Fungal effector proteins. Annu Rev Poppe S, Dorsheimer L, Happel P, Stukenbrock EH. 2015. Rapidly evolving Phytopathol. 47:233–263. genes are key players in host specialization and virulence of the fungal Stukenbrock EH, et al. 2011. The making of a new pathogen: insights wheat pathogen Zymoseptoria tritici (Mycosphaerella graminicola). from comparative population genomics of the domesticated wheat PLoS Pathog. 11(7):e1005055. pathogen Mycosphaerella graminicola and its wild sister species. Que Y, et al. 2014. Genome sequencing of Sporisorium scitamineum Genome Res. 21(12):2157–2166. provides insights into the pathogenic mechanisms of sugarcane Taniguti LM, et al. 2015. Complete genome sequence of Sporisorium smut. BMC Genomics 15:996. scitamineum and biotrophic interaction transcriptome with sugarcane. Raffaele S, Kamoun S. 2012. Genome evolution in filamentous plant PLoS One 10(6):e0129318. pathogens: why bigger can be better. Nat Rev Microbiol. Tellier A, Moreno-Gamez S, Stephan W. 2014. Speed of adaptation and 10(6):417–430. genomic footprints of host-parasite coevolution under arms race and Ranwez V, Harispe S, Delsuc F, Douzery EJP, Murphy WJ. 2011. MACSE: trench warefare dynamics. Evolution 68(8):2211–2224. multiple Alignment of Coding SEquences accounting for frameshifts Thompson JD, Plewniak F, Poch O. 1999. A comprehensive comparison of and stop codons. PLoS One 6(9):e22594. multiple sequence alignment programs. Nucleic Acids Res. Remy W, Taylor TN, Hass H, Kerp H. 1994. Four hundred-million-year-old 27(13):2682–2690. vesicular arbuscular mycorrhizae. Proc Natl Acad Sci U S A. Thorne JL, Kishino H, Painter IS. 1998. Estimating the rate of evolution of 91(25):11841–11843. the rate of molecular evolution. Mol Biol Evol. 15(12):1647–1657. Romiguier J, et al. 2012. Fast and robust characterization of time- Tiffin P, Moeller DA. 2006. Molecular evolution of plant immune system heterogeneous sequence evolutionary processes using substitution genes. Trends Genet. 22(12):662–670. mapping. PLoS One 7(3):e33852. Torun ~o TY, Stergiopoulos I, Coaker G. 2016. Plant-pathogen effectors: Rouxel T, et al. 2011. Effector diversification within compartments of the cellular probes interfering with plant defenses in spatial and temporal Leptosphaeria maculans genome affected by Repeat-Induced Point manners. Annu Rev Phytopathol. 54:419–441. mutations. Nat Commun. 2:202. Wicker T, et al. 2013. The wheat powdery mildew genome shows the Rovenich H, Boshoven JC, Thomma BPHJ. 2014. Filamentous pathogen unique evolution of an obligate biotroph. Nat Genet. effector functions: of pathogens, hosts and microbiomes. Curr Opin 45(9):1092–1096. Plant Biol 20:96–103. Yang Z. 1998. Likelihood ratio tests for detecting positive selection and Sainudiin R, et al. 2005. Detecting site-specific physicochemical selective application to primate lysozyme evolution. Mol Biol Evol. pressures: applications to the Class I HLA of the human major 15(5):568–573. 644 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Yang Z. 2006. Computational molecular evolution. Oxford Series in Zhan J, McDonald BA. 2013. Experimental measures of pathogen com- Ecology and Evolution, USA: Oxford University Press. petition and relative fitness. Annu Rev Phytopathol. 51:131–153. Yang Z. 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol Zuther K, et al. 2012. Host specificity of Sporisorium reilianum is tightly Biol Evol. 24(8):1586–1591. linked to generation of the phytoalexin luteolinidin by Sorghum bi- Yang Z, Nielsen R. 1998. Synonymous and nonsynonymous rate color. Mol Plant Microbe Interact. 25(9):1230–1237. variation in nuclear genes of mammals. J Mol Evol. 46(4):409–418. Associate editor: Rebecca Zufall Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 645 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Genome Biology and Evolution Oxford University Press

Positively Selected Effector Genes and Their Contribution to Virulence in the Smut Fungus Sporisorium reilianum

Free
17 pages

Loading next page...
 
/lp/ou_press/positively-selected-effector-genes-and-their-contribution-to-virulence-eWYaG1AbP0
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution.
ISSN
1759-6653
eISSN
1759-6653
D.O.I.
10.1093/gbe/evy023
Publisher site
See Article on Publisher Site

Abstract

Plants and fungi display a broad range of interactions in natural and agricultural ecosystems ranging from symbiosis to parasitism. These ecological interactions result in coevolution between genes belonging to different partners. A well- understood example is secreted fungal effector proteins and their host targets, which play an important role in pathogenic interactions. Biotrophic smut fungi (Basidiomycota) are well-suited to investigate the evolution of plant pathogens, because several reference genomes and genetic tools are available for these species. Here, we used the genomes of Sporisorium reilianum f. sp. zeae and S. reilianum f. sp. reilianum, two closely related formae speciales infecting maize and sorghum, respectively, together with the genomes of Ustilago hordei, Ustilago maydis,and Sporisorium scitamineum to identify and characterize genes displaying signatures of positive selection. We identified 154 gene families having under- gone positive selection during species divergence in at least one lineage, among which 77% were identified in the two investigated formae speciales of S. reilianum. Remarkably, only 29% of positively selected genes encode predicted secreted proteins. We assessed the contribution to virulence of nine of these candidate effector genes in S. reilianum f. sp. zeae by deleting individual genes, including a homologue of the effector gene pit2 previously characterized in U. maydis.Only the pit2 deletion mutant was found to be strongly reduced in virulence. Additional experiments are required to understand the molecular mechanisms underlying the selection forces acting on the other candidate effector genes, as well as the large fraction of positively selected genes encoding predicted cytoplasmic proteins. Key words: positive selection, effector evolution, smut fungi, comparative genomics, virulence. Introduction and fossil records (Remy et al. 1994; Gehrig et al. 1996; Plants and fungi have a long history of coevolution since the Martin et al. 2017). Different forms of plant–fungus interac- emergence of pioneering land plants 400 Ma. The develop- tions have evolved, including mutualistic symbiosis where ment of early plants was likely supported by associations with both plant and fungus benefit (Parniske 2008), and patho- symbiotic fungi, as suggested by analyses of ribosomal RNAs genic interactions where fungal colonization greatly reduces The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Genome Biol. Evol. 10(2):629–645. doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 629 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE plant fitness (Dean et al. 2012). Pathogenic interactions play Caryophyllaceae species (Aguileta et al. 2010), the wheat critical roles in natural and agricultural ecosystems, and un- pathogen Zymoseptoria tritici (Stukenbrock et al. 2011), the derstanding the evolutionary mechanisms shaping them is of rust fungus Melampsora larici-populina (Hacquard et al. great importance to plant production, food security, and pro- 2012), the rice BLAST fungus Magnaporthe oryzae (Huang tection of biodiversity in natural ecosystems (Fisher et al. et al. 2014), the wheat stem rust fungus Puccinia graminis 2012; Bagchi et al. 2014). f. sp. tritici (Sperschneider et al. 2014), the Irish potato famine Secreted fungal effector proteins are key players in patho- pathogen Phytophthora infestans (Dong et al. 2014), a group genic interactions as they are involved in protecting and of Fusarium species (Sperschneider et al. 2015), and a group shielding growing hyphae, suppressing plant defense of smut fungi parasitizing different grasses and a dicot host responses and changing plant physiology to support growth (Sharma et al. 2014, 2015). Yet, as only a few genes under of the pathogen (Stergiopoulos and de Wit 2009; de Jonge positive selection have been functionally studied, the link be- et al. 2011; Giraldo and Valent 2013). Many effector proteins tween the selected genotypes and their corresponding phe- lack known functional domains, and expression of a subset of notypes are only beginning to be understood and only a few effectors is linked to plant colonization (Lo Presti et al. 2015; studies used evolutionary predictions to unravel the molecular Toruno et al. 2016; Franceschetti et al. 2017; Lanver et al. mechanisms of host adaptation. For example, the population 2017). Effector proteins with a strong effect on virulence phe- genomics study in the wheat pathogen Z. tritici which identi- notype are thought to coevolve with their plant targets either fied candidate effector genes under positive selection in an arms race or a trench-warfare scenario (Brown and (Stukenbrock et al. 2011) was followed up experimentally, Tellier 2011). In the former, fungal effectors manipulating and in this case it was shown that the deletion of three of the host are under positive directional selection, and plant four candidate genes reduced virulence (Poppe et al. 2015). In targets evolve in response to changes in effector proteins the gray mold fungus Botrytis cinerea, four positively selected (Rovenich et al. 2014). In the latter scenario, sets of alleles genes were deleted without affecting virulence, and this find- are maintained by balancing selection in both host and path- ing was attributed to functional redundancy, the limited num- ogen populations (Brown and Tellier 2011; Tellier et al. 2014). ber of tested host plants, or experimental conditions different Several methods are available for identifying genomic regions from natural infections (Aguileta 2012). A study of the oomy- under selection (Nielsen 2005; Aguileta et al. 2009, 2010). It cete effector protein EpiC1 showed that a single amino acid has been proposed that genes with signatures of positive se- substitution at a site under positive selection affected the lection have important functions during host pathogen inter- binding affinity of different host proteases determining host action or have contributed to host specialization (Tiffin and specificity (Dong et al. 2014). Moeller 2006). It is therefore expected that the deletion of Smut fungi, belonging to the division of Basidiomycota, are such genes reduces virulence when tested on a susceptible a group of about 550 species parasitizing mostly grasses, in- host. cluding important crops like maize, sorghum, oat, barley, and Depending on the aim of the investigation, studies identi- sugarcane (Begerow et al. 2014). In smut fungi, sexual repro- fying genes with signatures of positive selection are carried duction is linked to pathogenic development and smut fungi out within or between species. Although studies on the pop- therefore depend on successful plant colonization to com- ulation level focus on recent and ongoing selective processes plete their life cycle. As biotrophic pathogens, they require and are instrumental in the understanding of adaptation, living plant tissue for establishing a successful interaction comparative genomic studies employing different species en- (Martinez-Espinoza et al. 2002). With few exceptions like compass a broader time span and provide insight into the Ustilago maydis, smut fungi usually develop symptoms only underlying genetic basis of host specialization (Plissonneau in the female or male inflorescence of their respective host et al. 2017). The signature of positive selection in such case plants. During the last 10 years, quality draft genome sequen- typically takes the form of an excess of divergence between ces of prominent species were obtained, including U. maydis, species due to increased fixation of mutations by selective thecausative agentofsmutdisease on maize and teosinte sweeps compared with a neutral expectation (Yang and (Kamper et al. 2006), Sporisorium reilianum causing head Nielsen 1998). This is commonly measured by the ratio of smut of maize and sorghum (Schirawski et al. 2010), nonsynonymous (d ) over synonymous (d ) divergence and Ustilago hordei infecting barley (Laurie et al. 2012), and N S a d /d ratio> 1 is taken as evidence for positive selection Sporisorium scitamineum parasitizing sugarcane (Queetal. N S under the assumption that synonymous substitutions are neu- 2014; Taniguti et al. 2015; Dutheil et al. 2016). The head tral while nonsynonymous are not. Positive selection studies in smut fungus S. reilianum occurs in two formae speciales a number of plant pathogen systems revealed that genes that infect maize (S. reilianum f. sp. zeae) or sorghum (S. encoding secreted effector proteins are enriched in signatures reilianum f. sp. reilianum)(Zuther et al. 2012). The concept of positive selection (Mo ¨ ller and Stukenbrock 2017). Such of formae speciales is used in phytopathology to distinguish studies include investigations in diverse plant pathogens like members of the same species based on their ability to colonize Microbotryum species causing anther-smut disease of a certain host plant (in this example maize or sorghum) 630 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE (Anikster 1984). The divergence of U. hordei, U. maydis, S. two members were aligned on the codon level using MACSE scitamineum,and S. reilianum wasinferredtohave occurred 1.01 b (Ranwez et al. 2011) and on the protein level using in the interval of seven and 50 Ma (Munkacsi et al. 2007). The PRANK v.100802 (Lo ¨ ytynoja and Goldman 2008). The result- availability of genome sequences of several species with dif- ing alignments were subsequently compared and column ferent host ranges, together with established tools for genetic scores (CS) computed for each position in the alignment manipulations (Brachmann et al. 2004; Kamper 2004; (Thompson et al. 1999). Only positions with CS of 100% Khrunyk et al. 2010; Schuster et al. 2016) make this group (that is, alignment columns identically found by both meth- of smut fungi particularly interesting to study the evolution of ods) and a maximum of 30% gaps were retained for further effector genes as well as their contributions to virulence, spe- analysis. ciation, and host specificity. Here, we employed the genome of the recently sequenced strain S. reilianum f. sp. reilianum SRS1_H2-8 (http://www.ebi. Estimation of Genome-Wide Divergence Values and ac.uk/ena/data/view/LT795054-LT795076)(Zuther et al. Divergence Times 2012) together with the genomes of U. maydis, U. hordei, S. scitamineum,and S. reilianum f. sp. zeae to identify poten- The five genomes of U. hordei, U. maydis, S. scitamineum, tial effector genes with signatures of positive selection. S. reilianum f. sp. zeae,and S. reilianum f. sp. reilianum Candidate genes were individually deleted in S. reilianum f. were aligned using the Multiz genome aligner from the sp. zeae and the phenotype of the deletion strains was TBA package (Blanchette et al. 2004) and projected on the assessed after infection of maize in order to understand their U. maydis genome as reference. The resulting multiple function with respect to virulence. We report that the deletion genome alignment had a total size of 21 Mb and was of one candidate gene, pit2, led to a strong reduction in vir- further restricted to regions with homologous sequences ulence and we further discuss hypotheses on the origin of in the five species (total length after this step: 14.3 Mb) positive selection for the other candidate genes. and processed to remove coding regions. The final non- coding sequence alignment had a total length of 2.2 Mb, for which pairwise nucleotide similarities were computed in nonoverlapping windows of 10 kb. Materials and Methods Gene families with exactly one member in each species Construction of Homologous Protein Families were concatenated and pairwise protein sequence similarities Fungal species used in this study, their number of gene mod- computed using the seqinr package for R (Charif and Lobry els, number of predicted secreted proteins, and sources of 2007). Protein alignments were also used to infer dates of genome data are listed in supplementary table S1, divergence, using a relaxed clock model. The PhyloBayes ver- Supplementary Material online. The predicted proteome of sion 4.1 software (Lartillot et al. 2009) was used with the the five smut fungi U. hordei, Umaydis, S. scitamineum, S. auto-correlated model of Thorne et al. (Thorne et al. 1998) reilianum f. sp. zeae,and S. reilianum f. sp. reilianum were under a GTRþ CAT model. A unique calibration point was used to perform an all-against-all BLASTp search (Altschul used, based on the divergence time of the most divergent et al. 1990). The SiLiX algorithm was subsequently used to lineage U. hordei, previously estimated to have occurred be- infer homology relationships based on the BLAST hits (Miele tween 27 and 21 Myr (Bakkeren and Kronstad 2007). A uni- et al. 2011). Two parameters are considered to decide form prior was used on this interval for the Monte-Carlo whether a BLAST hit can be taken as evidence for homology: Markov Chain. As convergence issues arise when large align- the percent identity between two sequences and the relative ments (>20,000 positions) are used, we followed the length of the hit compared with the total length of the two PhyloBayes authors’ recommendation to conduct a jackknife sequences, hereby referred to as “coverage.” In order to max- procedure. We generated three data sets of circa 20,000 imize the number of families comprising 1:1 orthologues (that amino acids by randomly sampling families and concatenating is families that have an equal number of members in each the corresponding alignments. Two chains were run in each species), SiLiX (Miele et al. 2011) was runwitha range for case and convergence was assessed. Sampling was per- coverage and identity thresholds between 5% and 95% in formed after a burning of 10,000 iterations, and every ten 5% steps. An identity of 40% and coverage between 5% and subsequent iterations. Chains were run to ensure that the 45% lead to the maximum number of families with 1:1 ortho- minimum effective sample size was >50 and maximum rela- logues (5,394; supplementary fig. S1, Supplementary Material tive difference <0.3 in at least one sample. Results are sum- online) while settings with 40% identity and 80% coverage marized in supplementary table S2, Supplementary Material lead to 5,326 families with 1:1 orthologues (supplementary online, and figure S2, Supplementary Material online, shows fig. S1, Supplementary Material online). Since using a higher the six chains for the three samples. In addition to the con- coverage had only a cost of 68 core families, the stricter cri- vergence of the two chains for each sample, our results reveal teria were applied for family clustering. Families with at least extremely consistent results between samples. Figure 1A Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 631 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE FIG.1.—Phylogeny, divergence estimates, and number of genes under positive selection in five related smut fungal species parasitizing different host plants. (A) Chronogram of the five fungal pathogens as estimated under a relaxed molecular clock. Boxes represent 95% posterior intervals, with corresponding values indicated below. (B) Pairwise sequence differences, for both the noncoding genome and the proteome (nonsynonymous differences). (C) Number of positively selected genes on each terminal branch (total number of genes and genes predicted to encode a secreted protein). shows estimates from one chain of the third data set, which positive selection for each gene family. This is typically shows a minimum effective sample size >300. achieved by measuring the ratio of nonsynonymous versus synonymous substitutions (d /d ratio) using models of codon N S sequence evolution (Yang 2006). In particular, nonhomoge- Detection of Positive Selection neous models of sequence evolution estimate the d /d ratio N S For gene families with at least three members, translated independently in different lineages, yet at the cost of potential sequences were employed to create maximum likelihood phy- overparametrization issues. In the manual of the PAML pack- logenetic trees using PhyML 3.0 (Guindon et al. 2010)with a age, the authors state that such models should only be used minimum parsimony starting tree and the LG amino acid sub- for hypothesis testing and advise against using them for scans stitution model with a four-classes gamma distribution of site- of positive selection. Dutheil et al. (2012) proposed a model specific substitution rate (Le and Gascuel 2008). The best tree selection approach (implemented in the TestNH package) topology obtained from nearest neighbor interchange (NNI) allowing to select for the best nonhomogeneous model sup- and subtree pruning recrafting (SPR) searches was kept ported by the data. They start by fitting the simplest (homo- (Guindon et al. 2010). BppML (Dutheil and Boussau 2008) geneous) model and sequentially add parameters to model was then used to re-estimate branch lengths from the codon variation of selective regime among lineages. Because the alignment using the YN98 substitution model (Nielsen and number of possible models is large even for small data sets, Yang 1998). We next aimed at inferring the occurrence of two heuristic approaches have been introduced: the “free” 632 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE heuristic permits unconnected branches from the tree to pathogenic development (Kamper et al. 2006; Schirawski evolve under the same regime, whereas the “joint” heuristic et al. 2010). In total, these clusters contain 163 genes, where restricts model sharing to connected branches (see 100 reside in clusters without virulence phenotype and 63 Dutheil et al. 2012 for details). The choice of models to test reside in clusters with virulence phenotype upon deletion. is guided by statistics on the patterns of substitutions on the Both types of clusters contain each 32 genes for which a phylogenetic tree, an approach named substitution mapping d /d ratio could be determined (the missing genes are part N S (Romiguier et al. 2012). Apart from the model selection ap- of families that do not have at least three members and were proach, the underlying models of codon sequence evolution therefore not analyzed). The d /d ratios of all genes in clus- N S are identical to the one originally described by Yang ters were compared between clusters with and without viru- (Yang 1998; Yang and Nielsen 1998). Model selection was lence phenotype (Wilcoxon rank-sum test). performed with the TestNH software, which contains two programs: 1) MapNH (Romiguier et al. 2012) was used for Gene Ontology Terms Enrichment Analysis mapping substitutions on the previously inferred phylogenetic All proteins in S. reilianum f. sp. zeae,in S. reilianum f. sp. tree and 2) PartNH (Dutheil et al. 2012) was subsequently reilianum,and in U. hordei were considered for Gene employed to fit time nonhomogeneous models of codon sub- Ontology (GO) term enrichment analyses. GO terms were stitutions. PartNH uses the previously inferred substitution assigned using iprscan 1.1.0 (http://fgblab.org/runiprscan;de- maps in order to perform model comparisons and select a veloped by Michael R. Thon) which links GO information pro- nonhomogeneous model with minimal number of parame- vided by Interpro to each protein. In this way, 1,759 unique ters. Both methods “free” and “join” were compared in or- GO terms could be assigned to 4,130 proteins in S. reilianum der to scan for positive selection. Finally, putative secreted f. sp. zeae, 1,744 unique GO terms could be assigned to effector proteins were identified by predicting secretion using 4,124 proteins in S. reilianum f. sp. reilianum, and 1,757 SignalP 4.0 (Petersen et al. 2011) and proteins were consid- unique GO terms could be assigned to 3,922 proteins in U. ered as secreted if the program indicated the presence of a hordei (supplementary table S3, Supplementary Material on- signal peptide but no transmembrane domain. line). The Bioconductor package topGO (Alexaetal. 2006) To detect residues under positive selection in homologues wasthen usedtolink eachGO termtothe three major cat- of pit2, the branch-site model with Bayes Empirical Bayes egories “Cellular Component,” “Biological Process,” or (BEB) analysisasimplemented in PAML4(Yang 2007) was “Molecular Function.” Enrichment analysis was performed applied. We employed information about family composition, by computing P values for each GO term using Fisher’s classic alignment, and phylogeny as outlined above and defined test with parent–child correction (Grossmann et al. 2007). sr10529 and srs_10529 as foreground branches. A posterior Cytoplasmic proteins with and without signatures of positive probability threshold of> 95% was used for the BEB analysis. selection were compared for the three species separately, and differences were considered to be significant at the 5% level. Association of Positively Selected Genes with Repeats in U. hordei Strains and Growth Conditions We tested whether genes under positive selection are located The Escherichia coli derivative Top10 (Invitrogen, Karlsruhe, significantly closer to repetitive elements than average genes Germany) and the Saccharomyces cerevisiae strain BY4741 in thegenomeof U. hordei, which shows the highest content (MATa his3D1 leu2D met15D ura3D;Euroscarf,Frankfurt, of repetitive elements in the group of smut fungi investigated Germany; kindly provided by M. Bo ¨ lker, Marburg) were here. For this analysis, only a group of “uncharacterized in- used for cloning purposes. Sporisorium reilianum f. sp. zeae terspersed repeats” was investigated, because it was shown strains used in this study are listed in supplementary table S4, previously that this is the only category showing a strong as- Supplementary Material online. They are derivatives of the sociation with candidate effector genes (Dutheil et al. 2016). haploid solopathogenic strain JS161 which is capable of plant Binary logistic regressions were conducted in R using the rms colonization without the need of a mating partner, because it package (Harrell 2015). The “robcov” function of the rms expresses a compatible pheromone/receptor pair (Schirawski package was used in order to get robust estimates of each et al. 2010). Escherichia coli was grown in dYT liquid medium effect. The variable “distance to the closest interspersed (1.6% [w/v] Trypton, 1.0% [w/v] Yeast Extract [Difco (R)], repeat” was transformed by log(xþ 1) because of its extreme 0.5% [w/v] NaCl), or YT solid medium (0.8% [w/v] Trypton, distribution. 0.5% [w/v] Yeast Extract, 0.5% [w/v] NaCl, 1.3% [w/v] agar) supplemented with 100 mg/mL Ampicillin when needed. The Comparing d /d Ratios of Genes Residing in Virulence N S yeast S. cerevisiae was maintained in YPD solid medium (1% Clusters [w/v] yeast extract, 2% [w/v] (R)-Pepton, 2% [w/v] (R)-Agar, Previous work has identified several virulence gene clusters in 2% [w/v] glucose) and grown on SC URA medium (1.7% U. maydis and some of them play important roles during [w/v] Yeast Nitrogen Base without ammonium sulfate, Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 633 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE 0.147% [w/v] dropout-mix without Uracil, 2% [w/v] glucose) The drag and drop cloning method in yeast (Jansen et al. for selecting transformants containing the plasmid pRS426 2005) was used to generate plasmids pRS426 Dsr12968 Hyg (Sikorski and Hieter 1989) (kindly provided by M. Bo¨lker, #1, pRS426 Dsr14944 Hyg #2, pRS426 Dsr10059 Hyg #1, Marburg) or derivatives of pRS426. Strains of S. reilianum pRS426 Dsr10182 Hyg #1, pRS426 Dsr14558 Hyg #1, and were grown in liquid YEPS medium (1.0% [w/v] yeast ex- pRS426 Dsr12897 Hyg #1 which contain deletion constructs light tract, 0.4% [w/v] peptone, 0.4% [w/v] sucrose) at 28 Con a for deleting the candidate genes sr12968, sr14944, sr10059, rotary shaker at 200 rpm. sr10182, sr14558,or sr12897. These plasmids are a derivate of plasmid pRS426, which can be maintained in E. coli and S. cerevisiae (Sikorski and Hieter 1989). PCR-amplified left and right borders of each candidate gene and the hygromycin Construction of S. reilianum Strains resistance cassette were integrated in pRS426 by homologous Polymerase chain reactions were performed using the Phusion recombination in S. cerevisiae. Subsequently, the hygromycin High-Fidelity DNA Polymerase (New England Biolabs). resistance cassette was replaced with the Geneticin resistance Templates were either JS161 genomic or indicated plasmid cassette by ligation via SfiI restriction sites, yielding plasmids DNA. Restriction enzymes were obtained from New England pRS426 Dsr12968 Gen #1, pRS426 Dsr14944 Gen #3, Biolabs. Protoplast-mediated transformation was used to pRS426 Dsr10059 Gen #1, pRS426 Dsr10182 Gen #1, transform S. reilianum following a method established for pRS426 Dsr14558 Gen #1, and pRS426 Dsr12897 Gen #5, U. maydis (Schulz et al. 1990). Transformants were selected respectively. Gene deletion constructs were PCR-amplified on RegAgar plates (1.0% [w/v] yeast extract, 0.4% [w/v] from the respective plasmid using listed primers (supplemen- Bacto(R)-Pepton, 0.4% [w/v] Sucrose, 1 M Sorbitol, 1.5% tary table S5, Supplementary Material online). The obtained [w/v] Bacto(R)-Agar) supplemented with 200 mg/mL deletion constructs were used to transform the S. reilianum Geneticin and true resistance was tested by growing single strain JS161 to generate the gene deletion strains colonies on PD plates (3.9% [w/v] Potato-Dextrose Agar, 1% JS161Dsr12968, JS161Dsr14944, JS161Dsr10059, [v/v] Tris–HCl [1 M, pH 8.0]) supplemented with 50 mg/mL JS161Dsr10182, JS161Dsr14558, and JS161Dsr12897, Geneticin. Gene replacements with resistance markers were respectively. generated with a PCR-based method employing the previ- The drag and drop cloning method was also used to gen- ously described SfiI insertion cassette system (Brachmann erate plasmid pRS426 Dsr12084 Gen #1. PCR-amplified left et al. 2004; Kamper 2004) and were confirmed by and right borders of sr12084 and the Geniticin resistance Southern blot analysis. Genomic regions residing 1-kb up- cassette were integrated in pRS426 by homologous recombi- stream (left border) or downstream (right border) adjacent to nation in S. cerevisiae. The gene deletion construct for delet- open reading frames of candidate genes were PCR-amplified ing the candidate gene sr12084 was PCR-amplified from using the listed primer pairs (supplementary table S5, plasmid pRS426 Dsr12084 Gen #1 using primers Supplementary Material online) and genomic DNA of JS161 sr12084_lb_fw/sr12084_rb_rv and transformed into the S. as template. The resulting fragments were used for cloning reilianum strain JS161 to generate the gene deletion strain plasmids containing the respective deletion constructs. JS161Dsr12084. To obtain deletion constructs for the genes sr10529 and sr14347, PCR fragments containing the left and right borders Virulence Assays of each gene were ligated to the hygromycin resistance cas- sette of pBS-hhn (Kamper 2004) via SfiI restriction sites and The solopathogenic strain JS161 and deletion mutants cloned into pCRII-TOPO (Life Technologies) to generate thereof were grown in YEPS liquid medium to an op- light pTOPO Dsr10529 #1 and pTOPO Dsr14347 #1, respectively. tical density at 600 nm (OD ) of 0.8–1.0 and cell cultures Since the use of Geneticin as selection marker resulted in were adjusted to an OD of 1.0 with sterile water prior much less false positive transformants compared with the to injection into 1-week-old maize (Zea mays) seedlings of use of Hygromycin B, the hygromycin resistance cassettes in the dwarf cultivar “Gaspe Flint” (kindly provided by B. these plasmids were replaced by the Geneticin resistance cas- Burr, Brookhaven National Laboratories and maintained sette of pUMA 1057 (Brachmann et al. 2004) by ligation via by self-pollination). Plants were sowed in T-type soil of SfiI restriction sites, yielding plasmids pTOPO Dsr10529 G418 “Fruhstorfer Pikiererde” (HAWITA, Vechta, Germany) and pTOPO Dsr14347 Gen #1, respectively. Deletion con- and grown in a temperature-controlled greenhouse structs were PCR-amplified from plasmids pTOPO Dsr10529 (14 h-/10 h-light/dark cycle, with 28/20 C and 25,000– G418 and pTOPO Dsr14347 Gen #1 using the listed primers 90,000 lux during the light period). Virulence symptoms (supplementary table S5, Supplementary Material online) and were scored 9- to 10-week postinfection according to used to transform the S. reilianum strain JS161 to generate previously described symptoms (Ghareeb et al. 2011) the gene deletion strains JS161Dsr10529 and and the following categories were distinguished: the plant JS161Dsr14347, respectively. did not develop ears, the plant developed healthy ears 634 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE shorter or equal to 1 cm or the plant developed healthy (supplementary table S6, Supplementary Material online). ears longer than 1 cm, the plant developed spiky ears, Gene duplications in U. hordei have been hypothesized to be phyllody in ears or phyllody in tassels. Spore formation driven by mobile elements (Laurie et al. 2012). was only observed occasionally, and rarely the plant died due to the infection. Three independent infections were The Genomes of S. reilianum f. sp. zeae and S. reilianum f. carried out per strain, and at least three independent de- sp. reilianum Diverged around 1 Ma letion strains were tested for virulence. To establish a frame for our comparative analysis, we first calculated sequence similarity of the five smut fungi for non- coding intergenic and protein sequences. In addition, we es- Results timated divergence times by performing a molecular dating Candidate Effector Genes Are Less Conserved between analysis based on the core orthologues set of the five patho- Species Compared with Other Genes gens, using advanced models of protein sequence evolution We reconstructed families of homologous genes for the five and Bayesian inference as implemented in the PhyloBayes smut fungi U. hordei, U. maydis, S. scitamineum, S. reilianum f. package (Lartillot et al. 2009). As calibration point, we used sp. zeae,and S. reilianum f. sp. reilianum using the SiLiX clus- thedivergencetimeof U. hordei and U. maydis, previously tering algorithm (Miele et al. 2011). We optimized the cluster- estimated to be between 27 and 21 Myr (Bakkeren and ing parameters to maximize the occurrence of orthologues and Kronstad 2007). In alignable intergenic regions U. hordei minimize the number of paralogues within each family. In this shares 57% identity with S. reilianum f. sp. reilianum and way, we were able to reconstruct 8,761 families, among which 77% identity in protein sequences (fig. 1B). Monte-Carlo 5,266 had at least one gene in each species (supplementary Markov chains were run for three independent gene samples table S6, Supplementary Material online). As a consequence, totaling >20,000 amino acid positions each, and two chains we found at least one homologous sequence in four species for were run in each case to assess convergence. The resulting 78% of all genes. About 5,254 gene families are found to have posterior distribution of divergence times were used to infer exactly one member in each species and were therefore taken 95% posterior intervals. The split between U. maydis and the as true orthologues (referred to as “core orthologous set” in Sporisorium species was estimated to have occurred 20 Myr the following). Considering that secreted proteins are putative ago (95% posterior interval 25–12 Myr; fig 1A and supple- effectors, we used SignalP (Petersen et al. 2011)to predict mentary table S2, Supplementary Material online). secretion of the encoded protein for each gene (supplementary Sporisorium reilianum f. sp. reilianum shares 61% nucleotide table S1, Supplementary Material online). We report that 920 identity in alignable intergenic regions with U. maydis,and (11%) families contained only genes encoding a predicted se- 79% sequence identity at the protein level (fig. 1B). The di- creted protein, whereas 7,657 (87%) contained only genes vergence times of S. scitamineum andthe twoformae spe- encoding a protein not predicted to be secreted. The remaining ciales of S. reilianum were calculated to be 13 Myr ago (95% 184 (2%) families contained both predicted secreted and cy- posterior interval 19–7 Myr; fig. 1A and supplementary table toplasmic proteins (supplementary table S6, Supplementary S2, Supplementary Material online), which is consistent with Material online). The occurrence of families with both secreted the mean divergence estimated between the hosts sorghum and cytoplasmic proteins can be explained by 1) false negative and sugarcane (10 Myr with a posterior interval of 8–13 Myr, predictions for secretion, as truncated C-terminal sequences average over eight studies, source: timetree.org; Kumaretal. were not removed from the data set, 2) wrong gene annota- 2017). Sporisorium reilianum f. sp. reilianum and S. scitami- tions, or 3) gain or loss of a secretion signal peptide during neum share 74% noncoding nucleotide identity and 88% effector evolution (Poppe et al. 2015). Among all predicted identity at the protein level (fig. 1B). Finally, the two S. reilia- secreted proteins, 52% have at least one orthologue in all other num strains diverged 1.1 Myr ago (95% posterior interval 2.4– species, which is significantly less than the global 78% propor- 0.4 Myr; fig. 1A) and share 98% noncoding nucleotide iden- 2 16 tion for all proteins (v test, P value< 2.210 ). Genes tity and 99% protein identity (fig. 1B). We note that the encoding putative effector proteins are therefore less conserved estimation of this divergence date varied with the gene set across species than other genes, either because their sequence used, and was in some cases found to be older (1.7 Myr, with is evolving faster, preventing the recovery of homologous rela- a 95% posterior interval of 4.7–0.6 Myr, see supplementary tionships, or because effector genes are created or lost at a table S2, Supplementary Material online). The comparison of higher rate. In U. hordei, we observe several species-specific the five smut genomes therefore encompasses a broad evolu- family expansions. There were 17 families which encompassed tionary time, and the divergence times obtained are compatible 5–25 members, but no orthologue in other species (supple- with previous estimates from smaller data sets (Munkacsi et al. mentary table S6, Supplementary Material online). Moreover, 2007). The speciation times of the investigated smut species we identified three families with up to 62 members in U. hor- largely predate the 10,000 years of crop plant domestication, dei, but only one member in up to three of the other species which implies that adaptation to the agricultural host, if any, Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 635 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE will be negligible when interpreting the interspecific patterns of of earlier studies in other pathosystems that showed that sequence divergence, as it represents a marginal proportion of predicted secreted proteins are often under positive selection, thetimesince thedivergencefrom the ancestral species. which can be attributed to their direct interaction with host proteins (Joly et al. 2010; Wicker et al. 2013; Poppe et al. 2015). Notably, all genes found under positive selection in Sporisorium reilianum Contains the Largest Number of the two strains of S. reilianum,in S. scitamineum and in U. Positively Selected Genes maydis share orthologous genes in the other species (supple- To detect positive selection, 6,205 families with at least mentary tables S3 and S6, Supplementary Material online). In three members (orthologues and/or paralogues, see supple- contrast, genes with signs of positive selection in U. hordei mentary table S6, Supplementary Material online) were, re- belong largely (36 out of 49 genes) to families showing gardless of their species composition, aligned on the codon species-specific expansions (supplementary tables S3 and and amino acid level and a phylogenetic tree was inferred. S6, Supplementary Material online). Obtaining accurate alignments is critical for detecting positive selection since alignment errors frequently inflate the false discovery rate (Schneider et al. 2009; Jordan and Goldman Genes Under Positive Selection in U. hordei Are Associated 2012). We therefore developed a stringent bioinformatics with Uncharacterized Interspersed Repeats pipeline for the filtering of sequence alignments by masking ambiguous alignment positions for further analysis (see Among the species compared here, the genome of U. hordei Materials and Methods). To scan for positive selection, we shows the highest fraction of repetitive elements (Laurie et al. employed a nonhomogeneous model of sequence evolution 2012; Dutheil et al. 2016). Such elements are known to con- allowing d /d ratios to vary along the phylogeny, in combi- tribute to gene family expansions (Kazazian 2004), and have N S nation with two heuristic model selection procedures to avoid been suggested to contribute to adaptation by providing ad- overparametrization issues (Nielsen and Yang 1998; Dutheil vantageous mutations, for instance by repeat-induced point et al. 2012). Model parameters could not be fitted by either mutations (RIP) leakage which was revealed in a species com- one of the two methods in 1.7% of branches. The two model plex of Leptosphaeria (Rouxel et al. 2011; Grandaubert et al. selection procedures led to highly consistent estimates of 2014). As sequence signatures of RIP were found in LTR ele- branch-specific d /d ratios (Spearman’s rank correlation co- ments of U. hordei (Laurie et al. 2012), we tested whether N S efficient equal to 0.85, P value< 2.210 ). The distribution genes under positive selection in U. hordei are physically as- of d /d was highly skewed with a median value of 0.06, sociated with repetitive elements. We performed a binary lo- N S demonstrating the strong predominance of purifying selec- gistic regression with the prediction of positive selection as a tion throughout lineages and genes. The mean value of d / response variable (that is, whether the underlying branch has d ratios for lineages undergoing positive selection (d /d >1) a d /d ratio >1) and we considered three putative explana- S N S N S was 4.1 (median 1.9). Although a d /d ratio >1 is indicative tory variables for each analyzed gene: 1) whether the gene is N S of positive selection, the absolute value of the ratio is a poor predicted to encode a secreted protein, 2) whether the gene is indicator of the strength of undergoing selection. In particular, duplicated, and 3) the distance of the gene to the closest high ratio values can be obtained because of low d values, interspersed repeat. The complete linear model explains and the d rate might include neutral substitutions, such as 50% of the observed variance, and the three explanatory nonsynonymous substitution that are conservative regarding variables are all significant at the 0.1% level (supplementary certain biochemical properties of the amino acids involved table S7, Supplementary Material online). These results sug- (Sainudiin et al. 2005). The largest number of genes with signs gest that positively selected genes in U. hordei are associated of positive selection was found in S. reilianum f. sp. zeae (84 with duplication events, and positive selection is more likely to genes, of which 25 encode predicted secreted proteins) and occur at genes encoding putative effectors. In addition, S. reilianum f. sp. reilianum (111 genes of which 27 encode the proximity of interspersed repeats increases the odds of predicted secreted proteins) (fig. 1C). In addition, a substantial positive selection, independently of the two other effects, number of positively selected candidate genes was also found and is confirmed by a stratification approach: the effect in U. hordei (49, and of these, 22 genes are predicted to code still holds when only duplicated genes are considered, or for secreted proteins), but only very few in U. maydis (2 genes) only genes encoding a secreted protein, or the combina- and S. scitamineum (7 genes) (fig. 1C). A list of all proteins tion of the two (supplementary table S7, Supplementary with their associated d /d ratios in each species is provided in Material online). This finding corroborates previous results N S supplementary table S3, Supplementary Material online. obtained in other microbial plant pathogens where it was Predicted secreted proteins were significantly enriched in described that effector genes tend to localize in repeat the group of proteins under positive selection in U. hordei rich regions and where it was suggested that such regions and in the two investigated formae speciales of S. reilianum contribute to the rapid evolution of effector genes (P values< 10 ; Fisher’s exact test). This corroborates results (Raffaele and Kamoun 2012). 636 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Table 1 Gene Ontology Terms Significantly Overrepresented in Positively Selected Genes Encoding Cytoplasmic Proteins in Sporisorium reilianum and Ustilago hordei a b Gene Ontology ID Gene Ontology Description Category P values Species GO: 0030532 Small nuclear ribonucleoprotein complex CC 0.023 Srr GO: 0045263 Proton-transporting ATP synthase complex, coupling factor F(o) CC 0.038 and 0.036 Srr and Srz GO: 0044425 Membrane part CC 0.047 Srr GO: 0031668 Cellular response to extracellular stimulus BP 0.015 and 0.013 Srr and Srz GO: 0051186 Cofactor metabolic process BP 0.015 Srr GO: 0042594 Response to starvation BP 0.017 and 0.012 Srr and Srz GO: 0009267 Cellular response to starvation BP 0.020 and 0.014 Srr and Srz GO: 0006790 Sulfur compound metabolic process BP 0.022 Srr GO: 0051301 Cell division BP 0.023 Srr GO: 0006777 Mo-molybdopterin cofactor biosynthetic process BP 0.027 Srr GO: 0009605 Response to external stimulus BP 0.027 and 0.012 Srr and Srz GO: 0043545 Molybdopterin cofactor metabolic process BP 0.034 Srr GO: 0043413 Macromolecule glycosylation BP 0.036 Srr GO: 0022402 Cell cycle process BP 0.037 Srr GO: 0051189 Prosthetic group metabolic process BP 0.039 Srr GO: 0006139 Nucleobase-containing compound metabolic process BP 0.012 Srz GO: 0016070 RNA metabolic process BP 0.023 Srz GO: 1901360 Organic cyclic compound metabolic process BP 0.016 Srz GO: 0006725 Cellular aromatic compound metabolic process BP 0.024 Srz GO: 0046483 Heterocycle metabolic process BP 0.023 Srz GO: 0035383 Thioester metabolic process BP 0.041 Srr GO: 1902589 Single-organism organelle organization BP 0.048 Srr GO: 0015074 DNA integration BP 0.039 Uh GO: 0006259 DNA metabolic process BP 0.023 Uh GO: 0010181 FMN binding MF 0.047 and 0.022 Srr and Srz GO: 0030515 snoRNA binding MF 0.021 Srz GO: 0016491 Oxidoreductase activity MF 0.028 Srz GO: 0016853 Isomerase activity MF 0.029 Uh CC, cellular component; BP, biological process; MF, molecular function. Calculated by Fisher’s classic test with Parent–Child correction. Only entries with P value 0.05 are shown. Species in which GO terms were found enriched. Srr, S. reilianum f. sp. reilianum;Srz, S. reilinaum f. sp. zeae;Uh, U. hordei. Positively Selected Genes Encoding Cytoplasmic Proteins in analysis for cytoplasmic proteins under positive selection in U. S. reilianum and U. hordei hordei was conducted, but only led to top-level categories (DNA integration, DNA metabolic process, and isomerase ac- Although we expect effector genes to be under positive se- tivity; see table 1). lection, we find that the majority of positively selected genes in S. reilianum encodescytoplasmic proteins(fig. 1C). To as- sess the putative functional role of these genes, we performed Virulence Contribution of Effector Genes Showing Signs of a Gene Ontology term enrichment analysis, comparing cyto- Positive Selection in S. reilianum plasmic proteins under positive selection to cytoplasmic pro- Candidate effector genes inferred to be under positive selec- teins not under positive selection (table 1). This analysis tion in a particular species could play a critical role in patho- revealed that genes with a potential role in metabolic pro- genicity. Therefore, we sought to assess the contribution to cesses, like sulfur compound metabolism, molybdopterin co- virulence of such candidate genes by creating individual dele- factor metabolic process, RNA metabolic process, organic tion mutants. In total, we tested nine candidate genes with cyclic compound metabolic process, and oxidoreductase ac- high d /d ratios and predicted to encode secreted proteins: N S tivity, as well as responses to starvation and extracellular stim- three with signatures of positive selection only in S. reilianum uli are significantly overrepresented at the 5% level (Fisher’s f. sp. zeae, three with signatures of positive selection in S. classic test with Parent–Child correction; see table 1). This reilianum f. sp. zeae as well as in S. reilianum f. sp. reilianum could indicate that cytoplasmic proteins under positive selec- and three with signatures of positive selection only in S. rei- tion contribute to metabolic changes which might be needed lianum f. sp. reilianum. All nine chosen candidate genes to- to survive with the limited nutrients available on the surface or gether with their characteristics are summarized in table 2. in the biotrophic interface of different host plants. A similar Deletion mutants were generated in the haploid Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 637 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE Table 2 Positively Selected Genes That Were Deleted in Sporisorium reilianum f. sp. zeae and Their Selection Criteria Family Gene ID Gene Description d /d Species with Number of Virulence Phenotype Closest Ortholog in N S Detected Positive Paralogs of Candidate Ustilago maydis a b Selection in Srz Gene Deletion FAM001428 sr10529 Conserved hypo- 31.1469 Srz and Srr 0 Virulence abolished UMAG_01375 (pit2) thetical protein FAM005472 sr10059 Conserved hypo- 6.53881 Srz and Srr 0 Virulence unaffected UMAG_05306 (cluster 19A) thetical Ustilaginaceae- specific protein FAM000532 sr10182 Conserved hypo- 1.57473 Srr 10 Virulence unaffected UMAG_00492 thetical protein FAM002067 sr12968 Conserved hypo- 37.9007 Srr 0 Virulence unaffected UMAG_02006 thetical protein FAM003728 sr14558 Conserved hypo- 24.355 Srz 0 Virulence unaffected UMAG_03564 thetical protein FAM004113 sr14944 Conserved hypo- 4.30527 Srz and Srr 0 Virulence unaffected UMAG_04034 (cluster 11–16) thetical Ustilaginaceae- specific protein FAM003465 sr14347 Conserved hypo- 544.37 Srz 5 Virulence unaffected UMAG_03349 thetical protein FAM001868 sr12897 Conserved hypo- 999 Srr 0 Virulence unaffected UMAG_01820 thetical protein e h k FAM000842 sr12084 Conserved hypo- 999 Srz 2 Virulence unaffected UMAG_00792 (cluster 1–32) thetical protein Species are S. reilianum f. sp. zeae (Srz) and S. reilianum f. sp. reilianum (Srr). Based on BLASTp search with an e-value cutoff of 0.001. BasedonBLASTpsearch. Maximum value allowed by the PartNH program. Infinity due to low value of d . The ten paralogs include: sr13431, sr11876, sr16607, sr11405, sr10621, sr16723, sr16877, 13293, sr11163.2,and sr15970. The five paralogs include: sr12257, sr11661, sr13976, sr14607,and sr11273. The two paralogs include: sr12085 and sr12086. As described in Doehlemann et al. (2011) and Mueller et al. (2013). As described in Ka €mperetal. (2006). As described in Schirawski et al. (2010). solopathogenic strain JS161 of S. reilianum f. sp. zeae. The gene sr10529 in S. reilianum f. sp. zeae is orthologous This strain is capable of colonizing maize plants and cause to the previously identified and characterized gene pit2 disease without a compatible mating partner (Schirawski (UMAG_01375)in U. maydis. et al. 2010) but virulence is much reduced relative to in- Pit2 plays an essential role in virulence as inhibitor of a fection with mating-compatible wild-type strains and group of maize papain-like cysteine proteases that are spores are only rarely produced. Deletion mutants were secreted to the apoplast (Doehlemann et al. 2011; also generated in strain JS161 in cases where positive se- Muelleretal. 2013). Previous work identified a conserved lection was only detected in S. reilianum f. sp. reilianum domain of 14 amino acids (PID14) in Pit2 as required and (table 2), because no solopathogenic strain is presently sufficient for the inhibition of maize cysteine proteases available for S. reilianum f. sp. reilianum. For each gene, (Muelleretal. 2013). When the branch-site model of at least three independent deletion mutants were gener- PAML 4 (Yang 2007) wasusedtoidentify amino acid ated and tested for virulence. To determine virulence, residues under positive selection in the Pit2 orthologues Gaspe Flint, a dwarf variety of corn, was infected and of the two S. reilianum species, only two residues residing symptoms were scored in male and female flowers in the PID14 domain were found under positive selection. (fig. 2). Only the deletion of sr10529, a gene showing However, 24 positively selected residues were detected positive selectioninbothformae speciales of S. reilianum, outside this domain in the 57 amino acid long C-terminal showed a strong reduction in virulence (table 2 and fig. 2). part (fig. 3). 638 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE n = 83 n = 88 n = 29 n = 29 JS161 #G3 #G4 #G5 123 4 JS161ΔSr10529 (Pit2) B F n = 95 n = 100 n = 95 n = 87 n = 93 n = 96 n = 95 n = 102 40 40 0 0 JS161 #3 #4 #5 JS161 #2 #3 #4 123 4 123 4 JS161ΔSr12968 JS161ΔSr14558 C G n = 95 n = 103 n = 103 n = 90 n = 83 n = 89 n = 85 n = 89 100 100 80 80 60 60 40 40 20 20 0 0 JS161 #15 #28 #31 JS161 #1 #2 #3 12 3 4 12 34 JS161ΔSr14944 JS161ΔSr14347 D H n = 95 n = 100 n = 95 n = 102 n = 95 n = 103 n = 103 n = 90 100 100 80 80 60 60 40 40 20 20 0 0 JS161 #25 #40 #47 JS161 #1 #2 #3 123 4 123 4 JS161ΔSr10059 JS161ΔSr12897 E I n = 92 n = 94 n = 93 n = 91 n = 94 n = 77 n = 94 n = 87 n = 82 20 20 0 0 JS161 #1 #2 #5 JS161 #4 #6 #9 #11 1234 12 3 4 5 JS161ΔSr10182 JS161ΔSr12084 plant without ears healthy ears ≤ 1 cm healthy ears > 1 cm spiky ears phyllody in ears ≤ 1 cm phyllody in ears > 1 cm phyllody in tassels spore formation dead plants FIG.2.—Virulence phenotypes of single deletion mutants of positively selected genes in Sporisorium reilianum f. sp. zeae. Plants of the maize variety “Gaspe Flint” were infected with the solopathogenic strain JS161 or independent deletion mutants of candidate genes as indicated below each bar. Deletion Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 639 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Symptoms of infected seedlings [%] Schweizer et al. GBE FIG.3.—Distribution of positively selected amino acids in the cysteine protease inhibitor Pit2. The alignment shows the protein sequences of orthologues in Ustilago hordei (UHOR_02064), U. maydis (UMAG_01375), Sporisorium scitamineum (SPSC_03677), S. reilianum f. sp. zeae (sr10529), and S. reilianum f. sp. reilianum (srs_10529). Sites under positive selection detected by a branch-site model are indicated by colored bold letters. Residues colored in red indicate positive selection detected in the respective species and purple residues indicate sites found under positive selection in both species. The yellow shaded area is orthologous to the previously identified conserved PID14 domain, which is required and sufficient for inhibition of a group of papain-like cysteine proteases. Green sequences indicate secretion signal peptides and bold numbers above the alignment indicate positions in UHOR_02064. positive selection over time spans of several millions of years Discussion is notoriously difficult (Gillespie 1994) because of two main We used evolutionary comparative genomics of five related reasons: 1) periods where genes are evolving under positive smut fungi infecting four different host plants to identify selection may occur episodically and may be followed by long genes with a signature of positive selection during species episodes of purifying selection, leading to an average d /d N S divergence, with a focus on genes encoding predicted se- <1 on long periods of time and 2) fast evolving genes may creted proteins, as such genes were suggested to contribute diverge to an extent where their homology is difficult to infer to virulence in various plant pathogenic microbes (Aguileta and where they can no longer be aligned reliably. To over- et al. 2010; Stukenbrock et al. 2011; Hacquard et al. 2012; come this problem, more genome information of species with Dong et al. 2014; Huang et al. 2014; Sharma et al. 2014).Our intermediate branching points is needed (Gillespie 1994). analysis revealed that positive selection is found between Predicted secreted proteins were about three times over- paralogous genes in U. hordei, where they belong to families represented in the set of positively selected genes, which with species-specific expansions. In contrast, genes under illustrates the importance of secreted proteins in adaptation positive selection in the other four species belong to families processes of smut fungi. This also corroborates results in other of orthologous sequences. Although we find evidence for a plant pathogenic microbes like Melampsora sp., Z. tritici and large set of genes under positive selection in the S. reilianum the wheat powdery mildew Blumeria graminis (Joly et al. species, signatures for positive selection are hardly detectable 2010; Stukenbrock et al. 2011; Wicker et al. 2013). in the more distant relatives U. hordei, U. maydis,and S. However, the majority of positively selected genes encodes scitamineum that diverged earlier. Finding evidence for FIG.2. Continued of sr10529 led toa strong reductioninvirulence (A). In contrast, deletion of the candidate genes sr12968 (B), sr14944 (C), sr10059 (D), sr10182 (E), sr14558 (F), sr14347 (G), sr12897 (H), and sr12084 (I) did not alter virulence. Symptoms were scored 9-week postinfection and categorized according to severeness as illustrated in the legend below the bar plots. Results are shown as mean of three independent experiments in relation to the total number of infected plants, which is indicated above each bar (n). Note that strains JS161DSr10529 #G4 and #G5 (A) were only infected in one replicate. 640 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE cytoplasmic proteins (fig. 1C), suggesting that both secreted due to the weak infection behavior of JS161. For example, and nonsecreted proteins are important targets of adaptation. spore formation is only rarely observed after infecting maize A Gene Ontology analysis in S. reilianum showed that mainly plants with the solopathogenic strain. In contrast, infections processes related to metabolism and its regulation as well as resulting from infections with two compatible haploid strains responses to starvation and external stimuli are enriched in show spores in 40% of the infected plants (Zuther et al. cytoplasmic proteins under positive selection. This points at a 2012). This means that defects related to spore formation will role of these proteins in adaptation to differences in nutrient not be evident in mutants of JS161. In three cases, positive availability in the respective host plants maize and sorghum as selection was detected in orthologous genes in S. reilianum f. well as responses to cues originating from the respective host sp. reilianum while candidate effector genes were for exper- (Haueisen and Stukenbrock 2016). A study conducted in U. imental reasons deleted in S. reilianum f. sp. zeae. Therefore, maydis has shown that the fungus induces major metabolic it cannot be excluded that these effectors might have a viru- changes in the host plant upon infection during establishment lence function in S. reilianum f. sp. reilianum. In this case, the of biotrophy and undergoes a series of developmental tran- positively selected effector genes might have evolved during sitions during host colonization that are likely influenced by adaptation to the sorghum host and present host specificity the host environment (Doehlemann et al. 2008). It is thus genes. In summary, our virulence assays leave open the pos- conceivable that the two S. reilianum accessions have adapted sibility that the eight candidate genes which did not show a to their different hosts that differ significantly for example in contribution to virulence could play a role in pathogenicity their amino acid and vitamin composition (Etuk et al. 2012). under conditions not tested here. Alternatively, candidate ef- Furthermore, recent studies in U. maydis suggested that in- fector proteins might also be positively selected for traits that tracellular changes of metabolism influence virulence, and are not directly linked to pathogenicity. Such traits could for therefore the underlying proteins could be targets of positive instance involve competition with large numbers of other selection (Kretschmer et al. 2012; Goulet and Saville 2017). plant colonizing microbes (Zhan and McDonald 2013; Out of nine deletions of positively selected genes, only one Rovenich et al. 2014). Secreted proteins of S. reilianum could mutant, lacking sr10529, was affected in virulence. Although act for example as toxin or could efficiently utilize resources six of the deleted genes are single genes in S. reilianum f. sp. from the environment and thereby limit the growth of other zeae for which we failed to identify paralogs, sr12084 has two microbes. In these cases, a contribution to virulence is not paralogs, sr14347 has five paralogs, and sr10182 has ten expected to be observed in the employed infection assay paralogs. We restricted our analyses to generating deletion with the effector gene mutants. Moreover, our molecular dat- mutants in some of the genes under positive selection. This ing analysis showed that the common ancestors of the inves- leaves open the possibility that the paralogous genes have tigated smut species originated before the beginning of crop redundant functions in virulence. Adapting the CRISPR-Cas9 domestication. Therefore, positive selection, whose signs we technology allowing multiplexing (Schuster et al. 2017)to S. detect by our approach, has most likely occurred on ancestral reilianum will be instrumental in testing this hypothesis in fu- host plants and not on the domesticated host maize. ture studies. Alternatively, the candidate effectors we investi- Consequently, some of the candidate effector genes under gated may be needed under conditions which differ from positive selection might not be important for the colonization those tested here. For example, S. reilianum f. sp. zeae can of crop plants, but for infection of related wild species. also systemically colonize maize plants via root infection In U. maydis, we note that effector genes residing in clus- (Mazaheri-Naeini et al. 2015), a colonization route we have ters whose deletion affected virulence (Kamper et al. 2006; not assessed in our experimental setup. Moreover, we Schirawski et al. 2010) have similar d /d ratios as effector N S employed only one maize cultivar for infection assays. genes in clusters where the deletion had no effect on viru- Results from other pathosystems suggest that virulence lence (median d /d ratio 0.0619 vs. 0.1094; Wilcoxon rank N S effects can strongly depend on the host and pathogen gen- test with P value¼ 0.1848). Furthermore, orthologues of otypes, in particular in the presence of R and avr genes (Petit- the effectors Pep1, Stp1, and Cmu1, which were shown Houdenot and Fudal 2017). No avr-R gene interaction was to have important roles in pathogenicity of U. maydis described so far in the S. reilianum f. sp. zeae-maize pathos- (Doehlemann et al. 2009; Schipper 2009; Djamei et al. ystem. Instead, quantitative virulence differences are observed 2011) showed no signature of positive selection. These when different host cultivars are infected (Lu ¨ bberstedt et al. observations could suggest that certain fungal effector 1999). Knowing the expression profile of effector genes may proteins are under evolutionary constraint and are there- assist the identification of differences in development of the fore not free to accumulate nonsynonymous mutations. mutants compared with wild-type strains. Since we lack this Such effectors are conserved over long time spans information, we scored disease symptoms only in the inflor- (Schirawski et al. 2010; Hemetsberger et al. 2015; escences 9 weeks after infection. Additionally, it may be Sharma et al. 2015) and this illustrates that they are in- possible that small differences in virulence between JS161 strumental for successful infections in a large group of and deletion mutants of candidate genes remain undetected smut fungi. They probably target molecules shared by Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 641 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE several host plants, for example, housekeeping functions of positively selected genes did not encode predicted secreted that cannot easily evolve in response to the binding of an proteins. Our results leave open the possibility that many genes effector. with signatures of positive selection contribute to virulence un- One candidate gene (sr10529) under positive selection in der conditions not tested in this study or are selected in traits both formae speciales of S. reilianum showed a strong con- that are not directly related to pathogenicity. tribution to virulence upon deletion. It is orthologous to the previously described protease inhibitor Pit2 in U. maydis, Supplementary Material where the deletion also abolished virulence (Doehlemann Supplementary data are available at Genome Biology and et al. 2011; Mueller et al. 2013). Positively selected residues Evolution online. in the PID14 domain of Pit2 might reflect that different pro- teases need to be inhibited in maize and sorghum. Pit2 might thus contribute to determining the host range of the respec- tive species. A role of cysteine protease inhibitors in host spe- cificity was demonstrated in Phytophthora infestans,a Acknowledgments pathogen of potato and its sister species P. mirabilis,which Our work was supported through the LOEWE program of the infects the ornamental plant Mirabilis jalapa. Positively se- state of Hesse through SYNMICRO and through the Max lected orthologous protease inhibitors were shown to inhibit Planck Society. G.S. was funded by the International Max proteases specific to the respective host plants and this spe- Planck Research School (IMPRS) for Molecular, Cellular and cificity could be traced back to a single amino acid substitution Environmental Microbiology. We thank Michael Bo¨lker for (Dong et al. 2014). Surprisingly, 24 positively selected sites in providing the yeast strain BY4741 and the plasmid pRS426 Pit2 were detected outside the PID14 domain in the 57 amino which were used for cloning purposes, and Benoit Nabholz acid long C-terminal part in both S. reilianum f. sp. zeae and S. and Nicolas Lartillot for guidance with the dating analysis. We reilianum f. sp. reilianum. This finding raises the intriguing are grateful to Markus Rampp (Max Planck Computing possibility that the C-terminus of Pit2 might possess a second Center Garching) for assistance with computational work function that is independent of protease inhibition. Earlier and to Stefan Schmidt for expert greenhouse service, as work has shown that the pit1 gene encoding a transmem- well as all present and former group members for lively dis- brane protein is located next to the pit2 effector gene and cussions and Eva H. Stukenbrock for her comments on the both genes contribute similarly to virulence (Doehlemann manuscript. et al. 2011). Furthermore, pit1 and pit2 are divergently tran- scribed, which makes it likely that the expression of pit1 and pit2 is coregulated. In addition, this gene arrangement of pit1 Literature Cited and pit2 is conserved in U. hordei, U. maydis, S. scitamineum, Aguileta G. 2012. Genes under positive selection in a model plant path- and S. reilianum (Sharma et al. 2015). This finding has led to ogenic fungus, Botrytis. Infect Genet Evol. 12(5):987–996. the speculation that Pit1 and Pit2 somehow act together to Aguileta G, et al. 2010. Finding candidate genes under positive selection in non-model species: examples of genes involved in host specialization govern virulence of U. maydis and related smut fungi. It was in pathogens. Mol Ecol. 19(2):292–306. hypothesized that that Pit2 shuttles apoplastic maize proteins Aguileta G, Refregier G, Yockteng R, Fournier E, Giraud T. 2009. Rapidly toward Pit1, thereby scavenging damage-associated mole- evolving genes in pathogens: methods for detecting positive selection cules (Doehlemann et al. 2011). In this scenario, the positively and examples among fungi, bacteria, viruses and protists. Infect Genet selected amino acids in the C-terminus of Pit2 could have Evol. 9(4):656–670. Alexa A, Rahnenfu ¨ hrer J, Lengauer T. 2006. Improved scoring of functional been selected for scavenging such molecules as adaptation groups from gene expression data by decorrelating GO graph struc- to the two hosts. In future studies, it will be highly interesting ture. Bioinformatics 22(13):1600–1607. to complement the pit2 mutant of S. reilianum f. sp. zeae with Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local the pit2 orthologue of S. reilianum f. sp. reilianum to seeifthis alignment search tool. J Mol Biol. 215(3):403–410. promotes virulence on sorghum. Anikster Y. 1984. The Formae Speciales. In: Bushnell WR, Roelfs AP, edi- tors. The cereal rusts. London: Academic Press. Bagchi R, et al. 2014. Pathogens and insect herbivores drive rainforest plant diversity and composition. Nature 506(7486):85–88. Conclusions Bakkeren G, Kronstad JW. 2007. Bipolar and Tetrapolar Mating Systems in Screens for genes with signs of positive selection are commonly the Ustilaginales. In: Heitman J, Kronstad JW, Taylor J, Casselton L, used to identify candidate effector genes in various plant path- editors. Sex in fungi. Washington: ASM Press. Begerow D, et al. 2014. Ustilaginomycotina. In: Esser K, editor. The ogenic microbes. However, it is currently largely open whether Mycota – a comprehensive treatise on fungi as experimental systems positively selected effector genes play indeed a role in virulence. for basic and applied research. Systematics and evolution (Part A). Here, we used comparative genomics of five smut fungi and Heidelberg: Springer. showed that only one out of nine genes under positive selection Blanchette M, et al. 2004. Aligning multiple genomic sequences with the contributes to virulence of S. reilianum. Moreover, the majority Threaded Blockset Aligner. Genome Res. 14(4):708–715. 642 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Brachmann A, Ko ¨ nig J, Julius C, Feldbru ¨ gge M. 2004. A reverse genetic Grossmann S, Bauer S, Robinson PN, Vingron M. 2007. Improved detec- approach for generating gene replacement mutants in Ustilago may- tion of overrepresentation of Gene-Ontology annotations with parent dis. Mol Genet Genomics 272(2):216–226. child analysis. Bioinformatics 23(22):3024–3031. Brown JKM, Tellier A. 2011. Plant-parasite coevolution: bridging the Guindon S, et al. 2010. New algorithms and methods to estimate gap between genetics and ecology. Annu Rev Phytopathol. maximum-likelihood phylogenies: assessing the performance of 49:345–367. PhyML 3.0. Syst Biol. 59(3):307–321. Charif D, Lobry JR. 2007. SeqinR 1.0-2: a contributed package to the R Hacquard S, et al. 2012. A Comprehensive Analysis of Genes Encoding project for statistical computing devoted to biological sequences re- Small Secreted Proteins Identifies Candidate Effectors in Melampsora trieval and analysis. In: Bastolla U, Porto M, Roman HE, Vendruscolo larici-populina (Poplar Leaf Rust). Mol Plant Microbe in 25(3):279–293. M, editors. Structural approaches to sequence evolution: molecules, Harrell FE Jr. 2015. Regression modeling strategies. Switzerland: Springer networks, populations. New York: Springer Verlag. p. 207–232. Series in Statistics. de Jonge R, Bolton MD, Thomma BP. 2011. How filamentous pathogens Haueisen J, Stukenbrock EH. 2016. Life cycle specialization of filamentous co-opt plants: the ins and outs of fungal effectors. Curr Opin Plant Biol. pathogens – colonization and reproduction in plant tissues. Curr Opin 14(4):400–406. Microbiol. 32:31–37. Dean R, et al. 2012. The top 10 fungal pathogens in molecular plant Hemetsberger C, et al. 2015. The fungal core effector Pep1 is con- pathology. Mol Plant Pathol. 13(4):414–430. served across smuts of dicots and monocots. New Phytol. Djamei A, et al. 2011. Metabolic priming by a secreted fungal effector. 206(3):1116–1126. Nature 478(7369):395–398. Huang J, Si W,Deng Q,Li P,Yang S.2014. Rapid evolution of Doehlemann G, et al. 2008. Reprogramming a maize plant: transcriptional avirulence genes in rice blast fungus Magnaporthe oryzae. and metabolic changes induced by the fungal biotroph Ustilago may- BMC Genet. 15:45. dis. Plant J. 56(2):181–195. Jansen G, Wu C, Schade B, Thomas DY, Whiteway M. 2005. Drag&Drop Doehlemann G, et al. 2009. Pep1, a secreted effector protein of Ustilago cloning in yeast. Gene 344:43–51. maydis, is required for successful invasion of plant cells. PLoS Pathog. Joly DL, Feau N, Tanguay P, Hamelin RC. 2010. Comparative analysis of 5(2):e1000290. secreted protein evolution using expressed sequence tags from four Doehlemann G, Reissmann S, Aßmann D, Fleckenstein M, Kahmann R. poplar leaf rusts (Melampsora spp.). BMC Genomics 11:422. 2011. Two linked genes encoding a secreted effector and a mem- Jordan G, Goldman N. 2012. The effects of alignment error and alignment brane protein are essential for Ustilago maydis-induced tumour for- filtering on the sitewise detection of positive selection. Mol Biol Evol. mation. Mol Microbiol. 81(3):751–766. 29(4):1125–1139. Dong S, et al. 2014. Effector specialization in a lineage of the Irish potato Kamper J. 2004. A PCR-based system for highly efficient generation of famine pathogen. Science 343(6170):552–555. gene replacement mutants in Ustilago maydis. Mol Genet Genomics Dutheil JY, Boussau B. 2008. Non-homogeneous models of sequence 271(1):103–110. evolution in the Bioþþ suite of libraries and programs. BMC Evol K€ amper J, et al. 2006. Insights from the genome of the biotrophic fungal Biol. 8:255. plant pathogen Ustilago maydis. Nature 444(7115):97–101. Dutheil JY, et al. 2012. Efficient selection of branch-specific models of Kazazian HH. 2004. Mobile elements: drivers of genome evolution. sequence evolution. Mol Biol Evol. 29(7):1861–1874. Science 303(5664):1626–1632. Dutheil JY, et al. 2016. A tale of genome compartmentalization: the evo- Khrunyk Y, Mu ¨ nch K, Schipper K, Lupas AN, Kahmann R. 2010. The use of lution of virulence clusters in smut fungi. Genome Biol Evol. FLP-mediated recombination for the functional analysis of an effector 8(3):681–704. gene family in the biotrophic smut fungus Ustilago maydis.New Etuk EB, et al. 2012. Nutrient composition and feeding value of Sorghum Phytol. 187(4):957–968. for livestock and poultry: a review. J Anim Sci Adv. 2:510–524. Kretschmer M, Klose J, Kronstad JW. 2012. Defects in mitochondrial and Fisher MC, et al. 2012. Emerging fungal threats to animal, plant and peroxisomal b-oxidation influcence virulence in the maize pathogen ecosystem health. Nature 484(7393):186–194. Ustilago maydis. Eukaryot Cell 11(8):1055–1066. Franceschetti M, et al. 2017. Effectors of filamentous plant pathogens: Kumar S, Stecher G, Suleski M, Hedges SB. 2017. TimeTree: a resource for commonalities amid diversity. Microbiol Mol Biol R timelines, timetrees, and divergence times. Mol Biol Evol. 81:e00066–e00016. 34(7):1812–1819. Gehrig H, Schussler A, Kluge M. 1996. Geosiphon pyriforme, a fungus Lanver D, et al. 2017. Ustilago maydis effectors and their impact on viru- forming endocytobiosis with Nostoc (cyanobacteria), is an ancestral lence. Nat Rev Microbiol. 15(7):409–421. member of the Glomales: evidence by SSU rRNA analysis. J Mol Lartillot N, Lepage T, Blanquart S. 2009. PhyloBayes 3: a Bayesian software Evol. 43(1):71–81. package for phylogenetic reconstruction and molecular dating. Ghareeb HA, Becker A, Iven T, Feussner I, Schirawski J. 2011. Sporisorium Bioinformatics 25(17):2286–2288. reilianum infection changes inflorescence and branching architectures Laurie JD, et al. 2012. Genome comparison of barley and maize smut fungi of maize. Plant Physiol. 156(4):2037–2052. reveals targeted loss of RNA silencing components and species-specific Gillespie JH. 1994. The causes of molecular evolution. Oxford: Oxford presence of transposable elements. Plant Cell 24(5):1733–1745. University Press. Le SQ, Gascuel O. 2008. An improved general amino acid replacement Giraldo MC, Valent B. 2013. Filamentous plant pathogen effectors in ac- matrix. Mol Biol Evol. 25(7):1307–1320. tion. Nat Rev Microbiol. 11(11):800–814. Lo Presti L, et al. 2015. Fungal effectors and plant susceptibility. Annu Rev Goulet KM, Saville BJ. 2017. Carbon acquisition and metabolic changes Plant Biol. 66:513–545. during fungal biotrophic plant pathogenesis: insights from Ustilago Lo ¨ ytynoja A, Goldman N. 2008. Phylogeny-aware gap placement prevents maydis. Can J Plant Pathol. 39(3):247–266. errors in sequence alignment and evolutionary analysis. Science Grandaubert J, et al. 2014. Transposable element-assisted evolution and 320(5883):1632–1635. adaptation to host plant within the Leptosphaeria maculans- Lu ¨ bberstedt T, Xia XC, Tan G, Liu X, Melchinger AE. 1999. QTL mapping Leptosphaeria biglobosa species complex of fungal pathogens. BMC of resistance to Sporisorium reiliana in maize. Theor Appl Genet. 99(3– Genomics 15:891. 4):593–598. Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 643 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Schweizer et al. GBE Martin FM, Uroz S, Barker DG. 2017. Ancestral alliances: plant mu- histocompatibility complex and the SRK of the plant sporophytic tualistic symbioses with fungi and bacteria. Science self-incompatibility system. J Mol Evol. 60(3):315–326. 356(6340):eaad4501. Schipper K. 2009. Charakterisierung eines Ustilago maydis Genclusters, Martinez-Espinoza AD, Garcia-Pedrajas MD, Gold SE. 2002. The das fu ¨ r drei neuartige sekretierte Effektoren kodiert. Philipps- Ustilaginales as plant pests and model systems. Fungal Genet Biol. Universitat Marburg, https://d-nb.info/1000245756/34. 35(1):1–20. Schirawski J, et al. 2010. Pathogenicity determinants in smut fungi Mazaheri-Naeini M, Sabbagh SK, Martinez Y, S ejalon-Delmas N, Roux C. revealed by genome comparison. Science 330(6010):1546–1548. 2015. Assessment of Ustilago maydis as a fungal model for root in- Schneider A, et al. 2009. Estimates of positive Darwinian selection are fection studies. Fungal Biol. 119(2–3):145–153. inflated by errors in sequencing, annotation, and alignment. Miele V, Penel S, Duret L. 2011. Ultra-fast sequence clustering from sim- Genome Biol Evol. 1:114–118. ilarity networks with SiLiX. BMC Bioinformatics 12:116. Schulz B, et al. 1990. The b alleles of U. maydis, whose combinations Mo ¨ ller M, Stukenbrock EH. 2017. Evolution and genome architecture in program pathogenic development, code for polypeptides containing fungal plant pathogens. Nat Rev Microbiol. 15(12):756–771. a homeodomain-related motif. Cell 60:295–306. Mueller AN, Ziemann S, Treitschke S, Aßmann D, Doehlemann G. 2013. Schuster M, Schweizer G, Kahmann R. 2017. Comparative analyses of Compatibility in the Ustilago maydis–maize interaction requires inhibi- secreted proteins in plant pathogenic smut fungi and related basidio- tion of host cysteine proteases by the fungal effector Pit2. PLoS mycetes. Fungal Genet Biol. 16:1087–1845. Pathog. 9(2):e1003177. Schuster M, Schweizer G, Reissmann S, Kahmann R. 2016. Genome edit- Munkacsi AB, Stoxen S, May G. 2007. Domestication of maize, sorghum, ing in Ustilago maydis using the CRISPR–Cas system. Fungal Genet and sugarcane did not drive the divergence of their smut pathogens. Biol. 89:3–9. Evolution 61(2):388–403. Sharma R, Mishra B, Runge F, Thines M. 2014. Gene loss rather than gene Nielsen R. 2005. Molecular signatures of natural selection. Annu Rev gain is associated with a host jump from monocots to dicots in the Genet. 39:197–218. smut fungus Melanopsichium pennsylvanicum. Genome Biol Evol. Nielsen R, Yang Z. 1998. Likelihood models for detecting positively se- 6(8):2034–2049. lected amino acid sites and applications to the HIV-1 envelope gene. Sharma R, Xia X, Riess K, Bauer R, Thines M. 2015. Comparative genomics Genetics 148(3):929–936. including the early-diverging smut fungus Ceraceosorus bombacis Parniske M. 2008. Arbuscular mycorrhiza: the mother of plant root endo- reveals signatures of parallel evolution within plant and animal patho- symbioses. Nat Rev Microbiol. 6(10):763–775. gens of fungi and oomycetes. Genome Biol Evol. 7(9):2781–2798. Petersen TN, Brunak S, von Heijne G, Nielsen H. 2011. SignalP 4.0: dis- Sikorski RS, Hieter P. 1989. A system of shuttle vectors and yeast host criminating signal peptides from transmembrane regions. Nat strains designed for efficient manipulation of DNA in Saccharomyces Methods 8(10):785–786. cerevisiae. Genetics 122:19–27. Petit-Houdenot Y, Fudal I. 2017. Complex interactions between fungal Sperschneider J, et al. 2014. Diversifying selection in the wheat stem rust avirulence genes and their corresponding plant resistance genes and fungus acts predominantly on pathogen-associated gene families and consequences for disease resistance management. Front Plant Sci. reveals candidate effectors. Front Plant Sci. 5:372. 8:1072. Sperschneider J, et al. 2015. Genome-wide analysis in three Fusarium Plissonneau C, et al. 2017. Using population and comparative genomics to pathogens identifies rapidly evolving chromosomes and genes associ- understand the genetic basis of effector-driven fungal pathogen evo- ated with pathogenicity. Genome Biol Evol. 7(6):1613–1627. lution. Front Plant Sci. 8:119. Stergiopoulos I, de Wit PJ. 2009. Fungal effector proteins. Annu Rev Poppe S, Dorsheimer L, Happel P, Stukenbrock EH. 2015. Rapidly evolving Phytopathol. 47:233–263. genes are key players in host specialization and virulence of the fungal Stukenbrock EH, et al. 2011. The making of a new pathogen: insights wheat pathogen Zymoseptoria tritici (Mycosphaerella graminicola). from comparative population genomics of the domesticated wheat PLoS Pathog. 11(7):e1005055. pathogen Mycosphaerella graminicola and its wild sister species. Que Y, et al. 2014. Genome sequencing of Sporisorium scitamineum Genome Res. 21(12):2157–2166. provides insights into the pathogenic mechanisms of sugarcane Taniguti LM, et al. 2015. Complete genome sequence of Sporisorium smut. BMC Genomics 15:996. scitamineum and biotrophic interaction transcriptome with sugarcane. Raffaele S, Kamoun S. 2012. Genome evolution in filamentous plant PLoS One 10(6):e0129318. pathogens: why bigger can be better. Nat Rev Microbiol. Tellier A, Moreno-Gamez S, Stephan W. 2014. Speed of adaptation and 10(6):417–430. genomic footprints of host-parasite coevolution under arms race and Ranwez V, Harispe S, Delsuc F, Douzery EJP, Murphy WJ. 2011. MACSE: trench warefare dynamics. Evolution 68(8):2211–2224. multiple Alignment of Coding SEquences accounting for frameshifts Thompson JD, Plewniak F, Poch O. 1999. A comprehensive comparison of and stop codons. PLoS One 6(9):e22594. multiple sequence alignment programs. Nucleic Acids Res. Remy W, Taylor TN, Hass H, Kerp H. 1994. Four hundred-million-year-old 27(13):2682–2690. vesicular arbuscular mycorrhizae. Proc Natl Acad Sci U S A. Thorne JL, Kishino H, Painter IS. 1998. Estimating the rate of evolution of 91(25):11841–11843. the rate of molecular evolution. Mol Biol Evol. 15(12):1647–1657. Romiguier J, et al. 2012. Fast and robust characterization of time- Tiffin P, Moeller DA. 2006. Molecular evolution of plant immune system heterogeneous sequence evolutionary processes using substitution genes. Trends Genet. 22(12):662–670. mapping. PLoS One 7(3):e33852. Torun ~o TY, Stergiopoulos I, Coaker G. 2016. Plant-pathogen effectors: Rouxel T, et al. 2011. Effector diversification within compartments of the cellular probes interfering with plant defenses in spatial and temporal Leptosphaeria maculans genome affected by Repeat-Induced Point manners. Annu Rev Phytopathol. 54:419–441. mutations. Nat Commun. 2:202. Wicker T, et al. 2013. The wheat powdery mildew genome shows the Rovenich H, Boshoven JC, Thomma BPHJ. 2014. Filamentous pathogen unique evolution of an obligate biotroph. Nat Genet. effector functions: of pathogens, hosts and microbiomes. Curr Opin 45(9):1092–1096. Plant Biol 20:96–103. Yang Z. 1998. Likelihood ratio tests for detecting positive selection and Sainudiin R, et al. 2005. Detecting site-specific physicochemical selective application to primate lysozyme evolution. Mol Biol Evol. pressures: applications to the Class I HLA of the human major 15(5):568–573. 644 Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Effector Genes and Their Contribution to Virulence in S. reilianum GBE Yang Z. 2006. Computational molecular evolution. Oxford Series in Zhan J, McDonald BA. 2013. Experimental measures of pathogen com- Ecology and Evolution, USA: Oxford University Press. petition and relative fitness. Annu Rev Phytopathol. 51:131–153. Yang Z. 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol Zuther K, et al. 2012. Host specificity of Sporisorium reilianum is tightly Biol Evol. 24(8):1586–1591. linked to generation of the phytoalexin luteolinidin by Sorghum bi- Yang Z, Nielsen R. 1998. Synonymous and nonsynonymous rate color. Mol Plant Microbe Interact. 25(9):1230–1237. variation in nuclear genes of mammals. J Mol Evol. 46(4):409–418. Associate editor: Rebecca Zufall Genome Biol. Evol. 10(2):629–645 doi:10.1093/gbe/evy023 Advance Access publication January 30, 2018 645 Downloaded from https://academic.oup.com/gbe/article-abstract/10/2/629/4830104 by Ed 'DeepDyve' Gillespie user on 16 March 2018

Journal

Genome Biology and EvolutionOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial