Portal de Periódicos da CAPES

The need for speed: compartmentalized genome evolution in filamentous phytopathogens

2018; Wiley; Volume: 20; Issue: 1 Linguagem: Inglês

10.1111/mpp.12738

1464-6722

Lamprinos Frantzeskakis, Stefan Kusch, Ralph Panstruga,

Legume Nitrogen Fixing Symbiosis

More than 10 years of extensive efforts since the dawn of high-throughput sequencing have resulted in a wealth of genome data for filamentous plant pathogens and have led to several key concepts in the research areas of genomics and plant-microbe interactions. Most recently, utilization of portable high-throughput sequencing devices for long-read sequencing has opened the door to new opportunities in pathogen population genomics and even for on-field pathogenomic investigations. We are, however, at a critical point where current genome evolution models are difficult to be reconciled with the plethora of different genome architectures and their supplementary characteristics. Initial pathogenomic analyses provided key insights on how the rapid evolution of some pathogens could be supported by plastic genomes (N.B.: In the context of this article, 'rapid' or 'fast' evolution is meant to indicate collectively all types of mutational events, i.e. nucleotide substitutions, insertions/deletions, rearrangements and copy number variation, that occur at a higher rate than 'normal', i.e. for example in housekeeping genes). Such plastic genomes are typically characterized by only moderately preserved synteny (gene collinearity) and frequent chromosomal and gene polymorphisms between isolates of the same species. One of the most widespread models in this respect is the 'two-speed' hypothesis, according to which large-scale genome compartmentalization is thought to accelerate, or at least be critical, for the evolution of genomic loci that include virulence-related genes (Dong et al., 2015). The two-speed concept is based on the idea that genomes of filamentous plant pathogens are compartmentalized and exhibit a mosaic genome architecture, comprising gene-dense/repeat-poor regions harboring essential and widely conserved housekeeping genes and gene-sparse/repeat-rich regions containing fast-evolving virulence-associated genes (e.g. effector genes) (Dong et al., 2015). Additional characteristics of two-speed genomes can be AT-rich isochores and accessory chromosomes, which altogether sustain genomic compartmentalization and are believed to allow for the rapid evolution of these genomes. Thus, the original two-speed hypothesis embraces two different aspects—(1) the presence of physical genomic compartmentalization and (2) the more rapid evolution of genes residing in one of these compartments. While the latter feature is clearly related to speed (i.e. the rate of mutational events per time unit), the former facet is a description of large-scale genomic organization and thus essentially unrelated to speed. In this sense, the phrase two-speed genome might be considered a misleading term, since the genome of every pathogen is assumed to have slowly evolving housekeeping and more rapidly evolving virulence-associated genes, i.e. having two speeds, regardless of its large-scale genomic organization. Since nevertheless this expression has been firmly established in scientific literature, we use it in the context of this article, but also propose an in our view more appropriate alternative (see below). The genomes of the oomycete Phytophthora infestans, the causal agent of the potato late blight disease, and its close relatives represent the original prototypes of a two-speed arrangement (reviewed in Dong et al., 2015). In these genomes, fast-evolving genes coding for effector proteins predominantly reside in gene-sparse regions, as exemplified by large intergenic distances within these compartments (Fig. 1A). In other prominent cases of the two-speed conformation, as the fungal pathogens Verticillium dahliae (causing Verticillium wilt; Faino et al., 2016) and Leptosphaeria maculans (causing blackleg disease; Rouxel and Balesdent, 2017), genome compartmentalization is realized by different means. In the first case, lineage-specific regions that harbor effector genes flanked by transposable elements (TEs; transposons) generate very dynamic loci where gene loss or duplications occur (Faino et al., 2016), while in the second case, AT-rich isochores generated by TE invasions contain rapidly evolving virulence genes. Fast evolution of effectors in isochores is thought to be associated with leaky RIP (Repeat-Induced Point mutations), a molecular mechanism that otherwise controls TE activity by introducing cytosine-to-thymine (C-to-T) nucleotide exchanges in repetitive sequences. Updated versions of two-speed genomes of filamentous pathogens supported with long-read sequencing data and/or extensive (re-)analyses solidify these initial observations and further suggest that virulence-associated regions of two-speed genomes evolve at a faster pace than the rest of the genome. Recent re-sequencing efforts, however, demonstrate that rapidly evolving phytopathogens that lack either some or all of the aforementioned characteristics exist, suggesting that there are more modes of genomic organization/evolution, possibly representing different 'speeds'. Here, drawing from the original two-speed hypothesis, we suggest and provide examples for two additional schemes: one-speed and multi-speed genomes. However, since 'speed' as an expression does not sufficiently cover the key aspect of large-scale genomic organization in these scenarios, we propose the adoption of the term 'compartment' (i.e. one-compartment genome, two-compartment genome etc.) instead in the future. For the purpose of this article, we will in the following mostly use both designations side-by-side. Updated genome assemblies based on long-read sequencing of the barley powdery mildew fungus Blumeria graminis f.sp. hordei (Frantzeskakis et al., 2018) and the wheat stripe rust fungus Puccinia striiformis f.sp. tritici (Schwessinger et al., 2018) indicate that in these cases, candidate secreted effector protein (CSEP)-coding genes are not predominantly found in gene-sparse areas as in other fungal pathogens. Additionally, these genomes do not show any obvious evidence of (large-scale) compartmentalization (Fig. 1). Especially for rust fungi, these recent studies further consolidate earlier results from the genome analysis of the poplar pathogen M. larici-populina and the aforementioned P. graminis f.sp. tritici. Since (rapid) evolution in these genomes is not constrained to certain compartments (which can be either TE-rich or not), we propose a 'one-speed' ('one-compartment') mode of evolution in these instances. In the suggested one-speed (one-compartment) species, (rapid) adaptation and evolution of virulence could be governed by copy-number variation (CNV) and heterozygosity of the effector loci. Unlike in some other fungi, CNV in these species is not linked to specific chromosomal regions (sub-telomeric/peri-centromeric). Even though CNV is over-represented for B. graminis CSEP genes, which are predicted to be involved in virulence, it can also be found for genes seemingly not directly associated with host interactions such as genes coding for serine/threonine kinase-like proteins, fungal transcription factors or hsp20 domain proteins (Frantzeskakis et al., 2018). Similarly, other phytopathogenic fungi that lack genomic compartmentalization, like the Armillaria species, present CNV in genes coding for a wider range of functions of which only some are potentially related to the infection process (Sipos et al., 2017). This suggests (1) that the mechanism(s) underlying these gene duplications or deletions act(s) across the whole genome, and (2) that in particular CNV in CSEP genes presumably provides direct advantages to virulence. For instance, avirulence effectors that exist in multiple copies or paralogs can accumulate non-synonymous mutations independently in each of the copies. This can result in complete or partial evasion of R gene-mediated resistance while presumably maintaining virulence function in the unaffected copy (Frantzeskakis et al., 2018). In a similar manner, it can be hypothesized that heterozygosity of effectors and the so far limited CNV that is reported in rusts can provide analogous selective advantages. Increased transcript dosage as a result of gene duplicates or multiple gene copies is believed to influence infection/survival success under selection, but may also provide space for transcriptional plasticity as well as sequence and functional diversification. Interestingly, while in two-speed genomes of ascomycetes RIP plays a crucial role in the evolution of effector sequences flanked by TEs, in powdery mildew fungi, it is the absence of RIP that allows widespread CNV unconstrained by gene size to exist (Frantzeskakis et al., 2018). Therefore, it can be hypothesized that the loss of RIP provides a balanced advantage against the over-proliferation of TEs in this case. Loss of RIP may inevitably result in a default genome architecture, at least as corroborated by the black truffle fungus Tuber melanosporum. This species is also devoid of RIP, which likely led to an expansion of TEs and inflation of the genome size up to ~4-fold bigger than that of its closest relative. In addition to the genome-wide distribution of TEs, CNV is observed in approximately 7% of the genome (Montanini et al., 2014). Despite the scarcity of additional cases, this indicates that the one-speed (one-compartment) genome architecture is not characteristic or unique to plant pathogens inhabiting highly selective environments, but also occurs in other, even plant-beneficial, eukaryotes. We propose that such one-speed (one-compartment) genomes are characterized by an over-proliferation of TEs due to the loss of genome defense mechanisms, equal distribution of TEs and absence of genomic compartmentalization. Rapid evolution in these genomes is supposedly enhanced by extensive CNV and/or heterozygosity of genes under selection. It cannot be ruled out, however, that such extant one speed (one-compartment) genomes evolved from an ancient two-speed (two-compartment) architecture, e.g. due to the loss of RIP and subsequent erosion of (previously existing) TE compartmentalization. In some other fungal pathogens, genome plasticity is governed predominantly by the acquisition or loss of accessory chromosomes. These accessory chromosomes carry in many cases rapidly evolving effector genes and are known to have a different epigenetic landscape than the rest of the genome. While the basic version of this genomic arrangement is captured by the two-speed (two-compartment) hypothesis, modified versions thereof have appeared in the recent literature. The wilt-causing pathogen Fusarium oxysporum seems to diverge from this established model of evolution by having a multi-speed (multi-compartment) genome (Fig. 1). In this case, dispensability expands to a core chromosome that is not necessary for virulence, which could potentially be creating three speed levels (compartments), including core, pseudo-core (i.e. core chromosomes lacking typical characteristics of accessory chromosomes but also displaying presence/absence polymorphism), and lineage-specific chromosomes (Vlaardingerbroek et al., 2016). At least in F. oxysporum, the accessory and pseudo-core chromosomes harbor duplication and/or deletion hotspots, potentially providing additional genome plasticity. This multi-speed (multi-compartment) hypothesis might gain more ground in the future as dense genomic data for more filamentous pathogens become available. Interestingly, in Zymoseptoria tritici not only loss, but also disomy (duplication) can be observed for accessory chromosomes (Fouché et al., 2018). In the same study it was reported that these two events happen at different frequencies for each accessory chromosome and are affected by their sequence composition, but also that disomy can be detected for core chromosomes as well. Along with the intrinsic characteristics of the dispensable chromosomes, extrinsic selection pressure is likely to affect also their loss, since they have been found to modulate virulence quantitatively. The scenarios outlined above demonstrate that while clearly an urgent 'need for speed' (= rapid evolution of virulence-associated genes) in genomes of filamentous phytopathogens is appreciated, the classical two-speed (two-compartment) genome represents just one realization of this necessity. Or, to state the argument the other way around, host selection pressure does not set pathogen genomes in a single evolutionary trajectory. Therefore, we propose that the two-speed (two-compartment) model cannot be generalized to include all rapidly evolving filamentous pathogens, and thus more categories or criteria have to be defined. At the end of the day, as in most aspects of biology, these groupings might not be strict, but borders between the categories are likely to be blurred and intermediate forms expected to exist. In fact, recent studies where previously sequenced genomes of filamentous phytopathogens were re-examined provide evidence of the diversity in genomic architectures. For example, the genome of the white mold fungus Sclerotinia sclerotiorum, despite its higher TE content compared to other members of the family, does not exhibit compartmentalization or other canonical two-speed (two compartment) features but shows a so far unique association of fast evolving virulence-related sRNAs with the location of TEs (Derbyshire et al., 2018). In the case of the genome of the rice blast fungus Magnaporthe oryzae, a recent long-read reassembly of the genome provides evidence for higher gene duplication rates and an approximately 10% higher TE content than previously reported (Bao et al., 2017). A divergence from the two-speed (two-compartment) model here is that secreted protein-coding genes are evenly distributed in the genome and that there is absence of recognizable large-scale genomic compartmentalization. Yet, a portion of these genes is in close proximity to TEs, which might lead to presence/absence polymorphisms and promote the potential gain of virulence in new hosts. However, there is no evidence that vicinity to TEs accelerates gene evolution (Okagaki et al., 2016). A final example comes from smut fungi (Ustilaginales), where in the genomes of some species effector genes are clustered, presumably due to their proximity to TEs, while in others clustering and TE insertion sites do not seem to correlate (Rabe et al., 2016). These studies indicate that even closely related species can follow different tracks of genomic evolution. The availability of more speed (compartment) categories could also help redefining the classification of some phytopathogen genomes. For example, in some cases genomes have been named two-speed although they fulfil either none or only a single of the aforementioned two-speed criteria. The genome of F. graminearum has not been found to have accessory chromosomes, has a low amount of TEs that do not generate compartmentalization, and AT-isochores are limited to centromeric/telomeric regions. Yet in recent reports it is classified as two-speed based on the uniqueness of its recombination patterns (Laurent et al., 2018). Similarly, the genome of the cereal net blotch pathogen Pyrenophora teres is categorized as two-speed (Wyatt et al., 2018) based on the presence of AT-isochores. However, it is still not clear how these isochores contribute to generation of virulence loci or if alternatively it is the activity of TEs that shape the genomes of the different formae speciales by disrupting synteny and effectively pseudogenize genes (Syme et al., 2018). As more (near-)complete high-quality genomes surface, we will likely be able to observe more clearly the full spectrum of genomic architectures realized for filamentous phytopathogens. This will also allow us to quantify how frequent each mode of genome evolution is, and perhaps we will gain a comprehensive explanation on how these genomic arrangements came to be and if they are indeed a result of selection pressure or rather a result of genetic drift and stochasticity. In addition, we have just begun to value in a limited number of species the extent to which chromatin biology influences the evolutionary speed of each genome irrespective of its large-scale organization. We envision that these multiple complementary hypotheses about the rapid evolution of filamentous phytopathogen genomes will fuel additional rounds of more thorough and standardized investigation on the architecture of chromosomes and stress the urgency to obtain more and complete genomic datasets. This work was supported by grant PA 861/14 within the Priority Programme SPP1819 'Rapid evolutionary adaptation: Potential and constraints' funded by the Deutsche Forschungsgemeinschaft (DFG). We thank Kaitlyn Courville for critical proofreading and apologize to all colleagues whose work could not be cited due to the strict guidelines for an 'Opinion' piece. We acknowledge the popular car racing video game 'Need for Speed' as a source of inspiration for the title of our article.