Are fungi‐derived genomic regions related to antagonism towards fungi in mosses?

Land plants have been intimately associated with fungi over the course of their evolution. Because of their lack of sophisticated protective structures, early land plants would conceivably have required additional defense strategies against microbial pathogens, including various fungi. On the other hand, a symbiotic association between plants and fungi is instrumental to plant adaptation to terrestrial environments (Selosse & Le Tacon, 1998; Bidartondo et al., 2011; Martin et al., 2017). The crucial role of this partnership is further evidenced by the widespread occurrence of mycorrhiza (root–fungi association) in vascular plants, as well as mycorrhiza-like fungal associations (MFAs hereafter) in nonvascular plants, such as liverworts and hornworts (Wang & Qiu, 2006; Pressel et al., 2014). Surprisingly, although fungal symbiosis is commonly considered to be an ancestral trait for land plants (Wang & Qiu, 2006; Delaux et al., 2015), with the possible exception of the genus Takakia (Boullard, 1988; Grosche et al., 2018), no MFAs have been confirmed in other mosses (Pressel et al., 2014; Field et al., 2015), the most diverse group of nonvascular land plants. Here, we report two genomic regions in the nuclear genome of the moss Physcomitrium patens, previously Physcomitrella patens (Medina et al., 2019; Rensing et al., 2020), that contain mostly fungi-specific genes and mobile genetic elements. These two regions were identified in our genome screening for horizontally acquired genes in P. patens. Available evidence indicates that these fungi-specific genes are probably involved in the interaction between mosses and fungi. We discuss how these fungi-specific genes might have contributed to the defense against fungal and other microbial pathogens, as well as the loss of MFAs in mosses. The two genomic regions reported here are 513 and 111 kbp long, respectively, and they are referred to as fungal region 1 (FR1) and fungal region 2 (FR2). FR1 is located on chromosome 13, from base pair position 14 225–527 422 bp, and contains 19 genes based on the annotation of P. patens v3.3 (Fig. 1a; Supporting Information Table S1) (Lang et al., 2018). With these genes as query, we performed Blast searches of NCBI nonredundant (nr) protein sequences, JGI Fungal Genomics (MycoCosm), OneKP (One Thousand Plant Transcriptomes, 2019), recently sequenced charophytes and hornworts, as well as our own genome and transcriptome data (Table S2; Methods S1). For eight of the 19 annotated genes, hits were mostly found in fungi and nonseed plants, including mosses, liverworts and seedless vascular plants (E-value threshold = 0.01); hits were also occasionally found in bacteria and microbial eukaryotes (e.g. haptophytes and stramenopiles), but not in charophyte algae, the closest relatives of land plants (Delwiche & Cooper, 2015), or in hornworts (Table S3). Based on their sequence similarities, these genes could be roughly classified into three families encoding: a heterokaryon incompatibility protein (HET) domain that reportedly functions in the self/nonself recognition system of filamentous fungi (PF06985; loci Pp3c13_190, Pp3c13_650/Pp3c13_670, and Pp3c13_930); a HopQ1-like protein (HLP) that may act as a virulence factor in bacteria (Pp3c13_910); and a conserved hypothetical protein with unknown function (Pp3c13_120, Pp3c13_300, and Pp3c13_690; termed PpCF here for P. patens conserved fungal gene (CF)) (Fig. 1a; Table S1). Another eight genes are either specific to P. patens or restricted to mosses and seedless vascular plants. The remaining three genes in FR1 are truncated but actively transcribed Ty1/Copia retrotransposons. FR2 contains 11 genes positioned from 5180 386 to 5291 530 bp on chromosome 27 (Fig. 1b; Table S1). Among them, seven genes from five loci contain a HET domain (Pp3c27_8610, Pp3c27_8690/Pp3c27_8695, Pp3c27_8710, Pp3c27_8740, Pp3c27_8810/ Pp3c27_8820), and one encodes a hypothetical protein of unknown function (Pp3c27_8730) that was only found in P. patens and two ascomycete fungi (Aspergillus turcosus, Fusarium euwallaceae) (E-value threshold = 0.01). Each of the remaining three annotated genes either encodes a short peptide (< 53 amino acids) or appears to be specific to P. patens. Both FR1 and FR2 contain additional degenerated transposable elements (TEs) (Fig. S1). The possibility that FR1 and FR2 represent a contamination could be excluded by the following evidence (Notes S1): CG, CHG and CHH DNA methylation (Fig. S1); homolog distribution of TEs from FR1 and FR2 in other chromosomes of P. patens, and in liverworts and lycophytes (Figs S2–S4); genetic linkage map and bacterial artificial chromosome/fosmid paired-end sequence data (Fig. S5); PCR amplification and sequencing of FR1, FR2, and adjacent regions (Fig. S6; Table S4); de novo sequencing and assembly of 12 other mosses, including one high-quality genome from an unspecified congeneric species (i.e. Physcomitrium sp.; Figs S7, S8). Other than a fraction of lineage-specific or orphan genes, all other genes in FR1 and FR2 were probably subjected to lateral transmission. Retrotransposons have long been known to transfer to, and amplify frequently in, plants and other eukaryotes (El Baidouri et al., 2014), and their presence in FR1 and FR2 is not surprising, particularly given that TEs account for 57% of the P. patens genome (Lang et al., 2018). The more noteworthy finding is the existence of three gene families with detectable homologs mostly restricted to nonseed plants and fungi (i.e. PpCF, HLP and HET domain-containing genes). Consistent with sequence similarity comparisons, phylogenetic analyses show that these genes in mosses and other land plants are most closely related to homologs from different fungal lineages (Figs 2, S9–S12; Methods S1; Notes S2). Such phyletic distributions and relationships are typically indicative of historical horizontal gene transfer (HGT) events between fungi and early land plants, although lineage-specific gene loss remains a possible, but less parsimonious, scenario (Huang & Gogarten, 2006) (also see Notes S2). Given their distribution in both Ascomycota and Basidiomycota (and Mucoromycota for CF), which originated earlier than land plants (Berbee et al., 2017; Lutzoni et al., 2018), we reason that these genes were probably transferred from fungi to the most recent common ancestor of land plants. In view of their affinities with different fungal lineages, the three gene families, and possibly members of HET domain-containing genes, might have been acquired independently and later evolved into gene clusters in P. patens through genome rearrangements; alternatively, they might have been derived from a single HGT event, followed by lineage-specific duplication and losses in both land plants and fungi. We note here that although genes acquired from various sources exist in land plants (Hoang et al., 2009; Yue et al., 2012; Maumus et al., 2014; Zhang et al., 2020), the vast majority of reported HGT events involve mitochondrial genes and/or parasitic plants (Davis & Xi, 2015; Yang et al., 2016); HGT from fungi to land plants was once considered to be extremely rare (Richards et al., 2009), though they have been documented in multiple recent studies (Bowman et al., 2017; Guan et al., 2018; Li et al., 2018; H. Wang et al., 2020; S. Wang et al., 2020). Our finding suggests that HGT of nuclear genes from fungi to plants might occur either more frequently than currently realized or in large fragments. Strikingly, all three gene families (i.e. CF, HLP and HET domain-containing genes) appear to be functionally related to fungal interactions with other organisms, either different fungi or host plants. HLPs contain an inosine-uridine nucleoside hydrolase domain and are homologous to the type III secretion system effector HopQ1 of bacterial pathogens (Li et al., 2013). The HLP in P. patens only shares 24–30% sequence identity with bacterial HopQ1, but it was shown to induce plant immune responses when ectopically expressed in the bacterium Pseudomonas syringae (Piechocki et al., 2018). This evidence suggests the HLP genes in fungi are probably involved in interactions with host plants. Indeed, the moss HLP protein sequences share the highest percentage identities (up to 41%) with plant-associated fungi such as the saprotroph Amorphotheca resinae, the leaf pathogen Marssonina coronaria, and the ericoid mycorrhizal fungus Meliniomyces variabilis. This pattern of gene distribution, with the highest similarities to sequences from fungal pathogens and symbionts, has also been observed for PpCF. Thus far, functional information for CF is not available, but gene fusion with an N-terminal HET domain was found in at least four homologous sequences from mycorrhizal fungi Pisolithus tinctorius and Pisolithus microcarpus (Fig. S13). As gene fusion often suggests functional linkage (Yanai et al., 2001), it is likely that CFs are functionally related to the HET domain in fungi. The HET domain is the most common component of heterokaryon incompatibility genes (het) in the vegetative self/nonself recognition system of filamentous fungi (Hall et al., 2010; Van der Nest et al., 2014). Such a system allows hyphal fusion for individuals identical at het loci, but induces programmed cell death (PCD) when hyphae with different het alleles encounter (Paoletti & Clave, 2007). Currently, the HET domain is believed to encode an effector for PCD in fungal heterokaryon incompatibility (Paoletti & Clave, 2007). In P. patens, HET domain-containing genes are also commonly distributed on other chromosomes; our analyses identified at least 27 HET domain-containing genes from 22 gene loci (Table S5; Fig. S14), most of which are combined with a C-terminal sequence region of unknown function that is also present in fungal homologs. The HET domain-containing genes, along with PpCF and HLP, are expressed in different tissues under various experimental conditions in P. patens (Figs S1, S14–S17). The majority of HET domain-containing genes have low expression levels. However, Pp3c14_10 and Pp3c27_8580 are moderately to strongly expressed in protonema, rhizoids, and gametophores (Figs S4, S151). Interestingly, Pp3c27_8710 and Pp3c12_12000 show more than two-fold stronger expression under exudate treatment with Gigaspora (Fig. S14), a genus of arbuscular mycorrhizal fungi. These genes are also found to be more highly expressed in quantitative reverse transcription polymerase chain reaction (qRT-PCR) under the treatment of chitooctaose, a chitin derivative (Galotto et al., 2020) (Fig. S16; Table S6). For the 16 genes for which data are present under both exudate and chitin treatments, we observe that 15 are more highly expressed under Rhizophagus exudate than in the control, while some are either less or more highly expressed under Gigaspora or Botrytis treatment (Fig. S18). The chitin treatment leads to genes either more highly expressed at 1 and 6 h, or showing higher expression after 1 h and lower expression after 6 h. Eight out of 16 genes show a congruent pattern of apparent expression stimulation under fungal treatment, while for the other eight genes there are differences between the treatments. This pattern of expression suggests a fine-grained response to different fungal treatments. We note that those loci encoding two genes (marked with blue boxes in Figs S14, S15) show a generally lower expression (Figs S14, S15), but they show detectable expression under treatment with both symbiotic and pathogenic fungal exudate (Fig. S14). The low level of detected expression might be a result of the challenge of representing conflicting gene models, sometimes on opposite strands. However, these regions encoding potential antisense RNAs might also constitute a defense system that makes use of small RNAs. The PpCF and HLP genes (Fig. S17) generally show stronger expression than the HET domain-containing genes. Similar to the HET domain-containing genes, three of the four PpCF and HLP genes show pronounced expression under fungal exudate (Fig. S17). The presence of the HET domain, PpCF, and other fungi-derived genes in mosses raises a fascinating question: why do mosses maintain genes that otherwise are specific to fungal activities? To date, most fungi-derived genes in mosses have not been investigated intensively, but it is worthwhile to speculate on their functional roles. It is possible that some of these genes have been coopted by mosses for activities not directly related to fungi. This is evidenced by the expression of Pp3c14_10 and Pp3c27_8580 during the various developmental stages of P. patens (Figs S14, S15). On the other hand, some other fungi-derived genes might have been recruited by mosses to regulate or counteract the activities of fungi or other microbes, in addition to their roles in other processes. This second possibility is in line with the role of the HLP gene in plant antimicrobial responses (Piechocki et al., 2018) and supported by similar cases documented in the literature (Chou et al., 2015; Di Lelio et al., 2019). For example, endogenous viral DNAs are known to contribute significantly to the immunity of host animals against viral infection (Aswad & Katzourakis, 2012). In P. patens, remnant sequences of giant viruses are thought to provide protection from nucleocytoplasmic viruses (Lang et al., 2018). Likewise, the Fusarium head blight resistance gene Fhb7 in wheatgrasses of the genus Thinopyrum was acquired from fungal endophytes (H. Wang et al., 2020). Other than the ectopic induction of plant immune response by HLP and the potential functional link between CF and the HET domain, additional lines of evidence are consistent with the possible role of these fungi-derived genes in moss defense. Importantly, het genes are known to transfer between species, and acquired het genes may indeed induce PCD in filamentous fungi (Paoletti et al., 2006; Wichmann et al., 2008). The het-c gene, which encodes a transmembrane protein, is one of the best-studied het genes in fungi. In Neurospora crassa, het-c forms the heterokaryon incompatibility system through nonallelic interactions with the closely linked pin-c gene, which contains the HET domain (Kaneko et al., 2006). Homologs of het-c, however, are also found in certain strains of the phytopathogenic bacterium Pseudomonas syringae, probably resulting from past HGT (Wichmann et al., 2008). Ectopic expression of bacterial het-c homologs in N. crassa induces PCD of hyphae, suggesting that P. syringae might have acquired het-c to kill fungal cells to obtain nutrients (Wichmann et al., 2008). We note here the other similarities of these genes in fungi and mosses. For instance, HET domain-containing genes are highly amplified and diverse in both groups. The HET domain has been proposed to evolve in fungi initially for pathogen recognition and host defense (Paoletti & Saupe, 2009). Interestingly, several HET domain-containing loci in P. patens (e.g. Pp3c13_930, Pp3c13_970) are located adjacent to loci encoding leucine-rich repeat receptor-like kinases (e.g., Pp3c13_960 and Pp3c13_990), major players in pathogen-associated molecular pattern recognition of plant immunity (Tena et al., 2011). Furthermore, several HET domain-containing genes are highly upregulated under the treatment of chitin, salicylic acid, and jasmonic acid, common signals of plant defense against pathogens and herbivores (Fig. S16). The acquisition of PpCF, HLP and HET domain-containing genes, which are seemingly specific to fungal activities, is surprising. The available evidence suggests that some of these genes might be involved in the interactions between mosses and fungi, probably as a defense mechanism in mosses. Such evidence is consistent with the idea that mosses and other bryophytes possess alternative defense strategies contributed by HGT from fungi and bacteria (Ponce de Leon & Montesano, 2017). These genes might have been lost secondarily from certain mosses but retained in other bryophytes or nonseed vascular plants, as often observed for resistance-related genes in plants (Zhao et al., 2018; Zhang et al., 2019). Given the likely antagonistic nature of these fungi-related genes, it is tempting to speculate whether they have possibly contributed to the loss of MFAs in mosses, particularly in light of the role of the HET domain in PCD of hyphae in filamentous fungi, which include mycorrhiza-like fungi. The loss of MFAs in mosses might also involve lineage-specific structural innovations and physiological processes (Field et al., 2015). Nonetheless, many questions remain to be answered. For instance, the HET domain in fungi contains three motifs of 15–30 amino acids with unknown functions (Paoletti & Clave, 2007). If some HET domain-containing genes are indeed involved in defense in mosses, further investigations are needed on whether all three motifs are required and whether these genes are intended for certain fungal groups or specific types of pathogens. Such information may help us to understand the existence of moss genes with a truncated HET domain (Table S5) and the presence of certain fungal pathogens and endophytes in mosses (Davey & Currah, 2006; U'Ren et al., 2010; Chen et al., 2018). Ultimately, gene knockout and other detailed functional investigations are required to uncover the role of these fungi-related genes in mosses. Such detailed functional investigations should also provide a better understanding of the importance of HGT in plant evolution, as well as in the interplay among genetic integration, organismal interaction and counteraction. We are grateful to John Stiller for reading the manuscript draft and for his constructive comments. Thanks are also due to Jianchao Ma, Guanxiao Chang, and Jinjie Zhao for assistance in data analysis and experimental verification of the studied genes. This work is supported in part by the National Science Foundation of China (NSFC) grants to JH (31970248), GS (31771414) and QW (31800190), the Department of Science and Technology of Henan Province to GS (18HASTIT041), the Chinese Academy of Sciences 'Lights of Western China' grant to JH, the Second Tibetan Plateau Scientific Expedition and Research (STEP) Program to HS (2019QZKK0502), and the Key Projects of the Joint Fund of the National Natural Science Foundation of China to HS (U1802232). The authors declare no competing financial interests. GS, YG and JH conceived and designed the study. GS, SB, YG, QW, JH, FBH, NF-P, AC and SAR performed analyses. SW performed experiments to confirm the existence and expression of identified genes. YL, HL, and BG generated additional genomic and transcriptomic data and contributed to data analyses. MP, YZ, XH and HS participated in data analyses. GS, SAR and JH wrote the manuscript. GS and SB contributed equally to this work. All sequences, alignments, phylogenetic trees, and scripts are available on the GitHub website (https://github.com/slbai01/2019-het-article). Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Fig. S1 JBrowse screenshots of the two regions covering FR1/2 in P. patens genome. Fig. S2 Molecular phylogeny of retroviral integrase core domain RVE (PF00665) type TEs in P. patens, Marchantia polymorpha and Selaginella moellendorffii and their chromosomal distribution in P. patens. Fig. S3 Molecular phylogeny of retroviral aspartyl protease RVP_2 domain (PF08284) in P. patens, M. polymorpha and S. moellendorffii and their chromosomal distribution in P. patens. Fig. S4 Molecular phylogeny of reverse transcriptase RVT_1 domain (PF00078) in P. patens, M. polymorpha and S. moellendorffii and their chromosomal distribution in P. patens. Fig. S5 Evidence of FR1 and FR2 in the genome assembly of P. patens. Fig. S6 Verification of regions in FR1 and FR2 by PCR amplification and sequencing. Fig. S7 Distribution of the target genomic regions (i.e. those mostly consisting of CF, HLP and HET domain-containing genes) in Physcomitrium sp., Ulota hutchinsiae and Anomodon attenuatus. Fig. S8 Mapping view and depth statistics of the contig that contains CF, HLP and HET domain-containing genes in Physcomitrium sp. (Contig00000005). Fig. S9 Maximum-likelihood tree of CFs in mosses and other eukaryotes. Fig. S10 Overview of the phylogenetic relationships of HET domain-containing genes in P. patens and other eukaryotes. Fig. S11 Molecular phylogeny of a subset of HET domain-containing proteins in P. patens and the homologs from other eukaryotes. Fig. S12 Molecular phylogenies of five HET domain-containing genes in FR1 and FR2. Fig. S13 Gene fusion of HET domain and CF homologs in fungi Pisolithus tinctorius and P. microcarpus. Fig. S14 Phylogenetic tree, expression profiles and domain structures of 27 HET domain-containing genes in P. patens. Fig. S15 Phylogenetic tree and expression profiles of 27 HET domain-containing genes in P. patens. Fig. S16 Relative expression levels of HET domain-containing genes in P. patens after exogenous application of SA, methyl jasmonate, and chitooctaose (a fungal elicitor). Fig. S17 Expression profiles of PpCF and HLP genes in P. patens. Fig. S18 Comparison of expression profiles of the HET domain-containing genes in P. patens. Notes S1 Evidence of the presence of FR1 and FR2 in P. patens. Notes S2 On the origin of HET domain-containing genes, HLP and CF in mosses. Methods S1 Identification of fungal genes in FR1, FR2, and their expression profiles. Table S1 Annotated genes in FR1 and FR2 of P. patens. Table S2 List of additional genomes and transcriptomes used in this study and identification of the three gene families. Table S3 Distribution of the HET domain-containing genes in nonseed land plants with genomic or transcriptomic data available in nr and OneKP databases. Table S4 Sequence-specific primers used for PCR verification of FR1, FR2, and their adjacent regions in P. patens. Table S5 Distribution of all the HET domain-containing genes in the genome of P. patens. Table S6 Sequence-specific primers used for qRT-PCR. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.