Dual Transcriptomics Reveals Interspecific Interactions between the Mycoparasite Calcarisporium cordycipiticola and Its Host Cordyceps militaris

White mildew disease caused by Calcarisporium cordycipiticola is devastating for the fruiting body cultivation of Cordyceps militaris, a popular and highly valued edible fungus. Here, the pathogenic mechanisms of C. cordycipiticola, the responses of C. militaris to the infection, and the interaction of these two phylogenetically close species were revealed by time course dual-transcriptome profiles. ABSTRACT Calcarisporium cordycipiticola is a mycoparasite of the edible fungus Cordyceps militaris, and mycoparasitism causes devastating diseases of mushrooms. In this study, dual-transcriptomic analysis was performed to reveal interspecific interactions between the mycoparasite C. cordycipiticola and its host C. militaris. At 4 and 8 days postinfection (dpi), 2,959 and 2,077 differentially expressed genes (DEGs) of C. cordycipiticola and 914 and 1,548 DEGs of C. militaris were identified compared with the mycelial stage, respectively, indicating that C. cordycipiticola responded more quickly than C. militaris. Lectins of the pathogen may play a role in the recognition of fungal prey. Both Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses showed that primary metabolism was vigorous for the pathogen to colonize the host and that the pathogen’s attack substantially altered C. militaris’ primary metabolism. C. cordycipiticola upregulated some carbohydrate-active enzyme (CAZyme) genes, including CBM18, GH18, GH16, and GH76, for degrading the host cell wall and defending against host immunity. C. militaris produced excessive reactive oxygen species (ROS) to respond to the infection. The GO term “heme binding” was the only shared term enriched at both stages at 4 and 8 dpi, indicating that iron was important for both the pathogen and the host. The uptake of iron by pathogens through multiple pathways promoted colonization and removed high ROS levels produced by the host. The transcription levels of Cmhsp78, Cmhsp70, and Cmhyd1 in C. militaris responded quickly, and these genes have potential as candidates for the breeding of resistant varieties. This study provides clues for understanding the interactions between a mycoparasite and its mushroom host and will be helpful for the breeding of resistant varieties and disease prevention and control for this edible fungus. IMPORTANCE White mildew disease caused by Calcarisporium cordycipiticola is devastating for the fruiting body cultivation of Cordyceps militaris, a popular and highly valued edible fungus. Here, the pathogenic mechanisms of C. cordycipiticola, the responses of C. militaris to the infection, and the interaction of these two phylogenetically close species were revealed by time course dual-transcriptome profiles. In general, the host C. militaris responds more slowly than the pathogen C. cordycipiticola. For the first time, we found that iron was important for both the mycoparasite and the host. C. cordycipiticola takes up iron by multiple pathways to promote colonization and remove high ROS levels produced by the host. The rapidly responding genes Cmhsp70, Cmhsp78, and Cmhyd1 in C. militaris may have the potential as candidate genes for the breeding of resistant varieties. This study expands our understanding of the mycoparasitic interactions of two species from sister families and will be helpful for the breeding of and disease prevention and control in mushrooms.

IMPORTANCE White mildew disease caused by Calcarisporium cordycipiticola is devastating for the fruiting body cultivation of Cordyceps militaris, a popular and highly valued edible fungus. Here, the pathogenic mechanisms of C. cordycipiticola, the responses of C. militaris to the infection, and the interaction of these two phylogenetically close species were revealed by time course dual-transcriptome profiles. In general, the host C. militaris responds more slowly than the pathogen C. cordycipiticola. For the first time, we found that iron was important for both the mycoparasite and the host. C. cordycipiticola takes up iron by multiple pathways to promote colonization and remove high ROS levels produced by the host. The rapidly responding genes Cmhsp70, Cmhsp78, and Cmhyd1 in C. militaris may have the potential as candidate genes for the breeding of resistant varieties. This study expands our understanding of the mycoparasitic interactions of two species from sister families and will be helpful for the breeding of and disease prevention and control in mushrooms. mycoparasite C. cordycipiticola and the host C. militaris. The genes differentially expressed genes during the infection processes were analyzed for C. cordycipiticola and C. militaris, and the interactions were then analyzed. The pathogenic mechanisms of C. cordycipiticola and the responses of C. militaris to the infection were revealed. This study will be helpful for the understanding of the interspecific interactions between this mycoparasite and its host and also for the breeding of resistant varieties of this mushroom.

RESULTS
Global transcriptome analysis during infection. To characterize the gene expression profiles of C. militaris and C. cordycipiticola during infection, time course dual-RNA-seq analyses were performed (see Fig. S1 in the supplemental material). The samples were harvested after being infected for 4 and 8 days and named 4 dpi (days postinfection) and 8 dpi, respectively. The mycelia of C. cordycipiticola (CC) and the healthy fruiting bodies of C. militaris (CM) used for infection served as the controls (Fig. 2).
Illumina paired-end sequencing generated a total of 151.76 Gb of clean data, with averages of 9.45 Gb per pure cultured sample and 15.8 Gb per infected sample (Table S1). All Q 30 percentages for the sequences (with an error probability of 0.01) in the 12 libraries  Interactions between C. cordycipiticola and Its Host Microbiology Spectrum were over 91.50% (Table S1). For the CC sample, 93.17% 6 0.40% of reads were uniquely mapped to the C. cordycipiticola reference genome, and only 0.27% 6 0.09% were mapped to that of the host C. militaris. Similarly, for the CM sample, 85.27% 6 1.37% of the reads were uniquely mapped to the C. militaris reference genome, and 1.34% 6 0.63% were mapped to that of the pathogen C. cordycipiticola (Table S1). The mapped data for each sample had genome coverages that were over or close to 100Â (Table S1). Although C. militaris and C. cordycipiticola are phylogenetically closely related, their transcriptome data can be completely separated, which lays the foundation for our interaction transcriptome analysis. Pearson's correlation coefficients of gene expression between replicates for each sample were .0.86, indicating good repeatability (Fig. S2).
Differentially expressed genes of Calcarisporium cordycipiticola during infection. For C. cordycipiticola, 2,959 and 2,077 differentially expressed genes (DEGs), which were 28.33% and 19.89% of the total genes, were identified at 4 and 8 dpi compared with CC, respectively ( Fig. 3A and B), and thus were involved in the interaction. Gene ontology (GO) analysis showed that the DEGs were enriched in 19 and 24 terms at 4 and 8 dpi, respectively (P , 0.01) (Fig. 4A). Among them, 5 terms were shared, and "structural constituent of ribosome" was the most enriched term at both stages (Fig. 4A). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed that there were 3 and 2 significant pathways at 4 and 8 dpi (corrected P value of ,0.05), respectively, and the ribosome pathway was the most significant pathway at both stages (Table S2). Both GO and KEGG analyses showed that the ribosome pathway was important, suggesting that protein synthesis was vigorous in the pathogen during infection. The KEGG pathways oxidative phosphorylation and propanoate metabolism were especially enriched (Table  S2) at 4 dpi. One KEGG pathway (starch and sucrose metabolism) was especially enriched . The red, yellow, and blue bars represent biological processes, cellular components, and molecular functions, respectively. Only significant GO terms (P , 0.01) are shown. Asterisks indicate GO terms with a P value of ,0.001. Shades of green indicate the shared GO terms at the two stages (4 and 8 dpi). ncRNA, noncoding RNA; FAD, flavin adenine dinucleotide; IHVP, interaction with host via protein; SAS, secondary active sulfate.
The DEG with the highest transcription level at 4 dpi was CCOR_05591, encoding a secreted small cysteine-rich protein with a common fungal extracellular membrane (CFEM) domain. It was upregulated 3-fold at 4 dpi, with an average fragments per kilobase of transcript per million fragments mapped (FPKM) value of 15,829 (Table S3). CFEM domain proteins are specific to fungi (18) and are involved in differentiation, stress responses, and pathogenicity (19). Eleven proteins with CFEM domains (CcCFEM1 to CcCFEM11) were identified in C. cordycipiticola, and most Cccfem genes were highly expressed during infection (Table S5).
Transcriptional response of Cordyceps militaris induced by infection with Calcarisporium cordycipiticola. For C. militaris, 914 and 1,548 DEGs, which were 9.47% and 16.04% of the total genes, were identified at 4 and 8 dpi compared with CM, respectively (Fig. 3C). There were many more DEGs at 8 dpi, indicating that C. militaris may respond slowly. The number of upregulated genes was higher than the number of downregulated genes at both 4 and 8 dpi (Fig. 3D).
GO analysis showed that DEGs were enriched for 13 and 15 terms at 4 and 8 dpi, respectively (P , 0.01) (Fig. 4B). Among them, 4 terms were shared, and "heme binding" was the most enriched term at both stages (Fig. 4B). KEGG analysis showed that 5 and 7 significant pathways were enriched at 4 and 8 dpi, respectively. 2-Oxocarboxylic acid metabolism was the most significant pathway at both stages (corrected P value of ,0.05) (Table S2). 2-Oxocarboxylic acids, also called 2oxo acids and a-keto acids, are the most elementary set of metabolites, which include pyruvate (2-oxopropanoate), 2-oxobutanoate, oxaloacetate (2-oxosuccinate), and 2-oxoglutarate (20). "Butanoate metabolism" was especially enriched at 4 dpi, and "biosynthesis of amino acids," "biosynthesis of antibiotics," and "arginine biosynthesis" were especially enriched at 8 dpi (Table S2). Both GO and KEGG analyses showed that the pathogen's attack substantially altered the primary metabolism of C. militaris.
Among all of the genes of C. militaris, the most upregulated gene was CCM_05892 encoding CmHSP78, which showed a 10.26-fold change at 4 dpi compared with CM (Table S3). Fifty-one heat shock protein (HSP)-encoding genes were identified in the genome of C. militaris (Table S6), among which 20 were expressed differently at 4 and 8 dpi. Most of the differentially expressed hsp genes (75%; 15/20) showed higher expression levels at 4 dpi than at 8 dpi, especially Cmhsp70 genes (Table S6). This was in sharp contrast to the fact that there were many more DEGs at 8 dpi than at 4 dpi for C. militaris (Fig. 3D), indicating that hsp genes are involved in the early response to infection. The upregulation of Cmhsp78 (CCM_05892) and Cmhsp70 (CCM_03788) after infection for 4 days was confirmed by quantitative real-time PCR (qPCR) (Fig. 5). Cmhsp78 and Cmhsp70 qPCR primers are listed in Table S7.
The DEG with the highest transcription level at 4 dpi was Cmhyd1 (CCM_03537), one of the hydrophobin genes. It was expressed with an average FPKM value of 11,334 and showed 5-fold upregulation at 4 dpi, which then decreased slightly at 8 dpi (Table S3). Four hydrophobin genes, termed Cmhyd1 to Cmhyd4, were identified in the C. militaris genome (21). There was no response of Cmhyd2 to Cmhyd4 at the infection stage, and they maintained low expression levels. The upregulation of Cmhyd1 was confirmed by qPCR ( Fig. 5). It was speculated that Cmhsp78, Cmhsp70, and Cmhyd1 were related to the defense response.
Genes predicted to be involved in recognition and sensing between mycoparasites and hosts. Genes involved in the recognition of fungal prey include those that encode G-protein-coupled receptors (GPCRs) and lectins (22,23). All 23 predicted GPCR genes in the genome were expressed in samples from the three stages. The genes CCOR_06769 (encoding a plasma membrane protein Pth11-like protein) and CCOR_05453 (encoding microbial opsins) showed the highest transcription levels at 4 dpi (Table S8). Fourteen lectin-related genes were screened in C. cordycipiticola, and six were upregulated at 4 dpi. Five of them showed the highest transcription levels at 4 dpi, including three encoding proteins with a ricin-type b-trefoil lectin domain (CCOR_01894, CCOR_03676, and CCOR_08831) and two encoding proteins with a legume-like lectin family domain (CCOR_01446 and CCOR_08418) ( Table S9).
The recognition of pathogen-associated molecular patterns (PAMPs) is a fundamental host survival mechanism (24). There have been almost no reports on the PAMPs of mushrooms. The transcripts of 18 genes encoding lectins in C. militaris were analyzed (Table S9), and it was found that there was no distinct differential expression at both 4 and 8 dpi compared with CM. Only the CCM_05447 gene encoding concanavalin A-like lectin and the CCM_07787 gene encoding mannose-binding lectin were upregulated slightly at 4 or 8 dpi, and two genes encoding proteins with a ricin-type b-trefoil lectin domain (CCM_03178 and CCM_03832) were downregulated slightly at 4 or 8 dpi.
Differential expression of CAZymes in Calcarisporium cordycipiticola and Cordyceps militaris. Among the 441 predicted genes encoding carbohydrate-active enzymes (CAZymes) in the genome of C. cordycipiticola (16), 213 (48%) were differentially expressed at 4 or 8 dpi compared with CC (Table S10). The genes were grouped into three main clusters according to their expression patterns (Fig. 6A). The genes of cluster I showed upregulation at both 4 and 8 dpi and occupied 53.52% (114/213) of the total differentially expressed CAZyme genes. There were 55 (25.82%; 55/213) genes in cluster II, and they were downregulated after infection. These CAZymes should be related to carbohydrate metabolism when C. cordycipitcola is saprophytic. The genes of cluster III (20.66%; 44/213) were expressed at the highest levels at 8 dpi, and they should respond slowly during infection ( Fig. 6A; Table S10).
The most upregulated CAZyme gene at 4 dpi was CCOR_03676, which had very low expression levels at CC and was upregulated by over 8-and 7-fold at 4 and 8 dpi compared with CC (Table S10). CCOR_03676 encodes a protein of glycoside hydrolase family 76 (GH76), which is an endo-acting a-mannanase for degrading a-1,6-manooligosaccharides or the a-1,6-linked backbone of fungal mannoproteins (25). Among the 10 predicted GH76encoding genes in the genome of C. cordycipiticola (16), 7 showed upregulation at 4 and Interactions between C. cordycipiticola and Its Host Microbiology Spectrum 8 dpi compared with CC (Table S10). It was reported previously that there was 9.0 to 17.5% mannose in the cell wall of the host C. militaris (26), indicating that the degradation of the cell wall of the host is important for infection by the pathogen C. cordycipiticola. The second and third most upregulated CAZyme genes at 4 dpi were CCOR_09360 (Cccbm18-7) and CCOR_06535 (Cccbm18-4), both belonging to carbohydrate-binding module 18 (CBM18) (Table S10). qPCR confirmed that Cccbm18-7 and Cccbm18-4 were highly expressed at 4 and 8 dpi, respectively (Fig. 6C). CBM18 is a subclass of chitin-binding domains that have convergently evolved in fungi, plants, and arthropods (27). There were 7 predicted proteins (CcCBM18-1 to CcCBM18-7) in the genome of C. cordycipiticola with CBM18 domains, which can be identified as Pfam domain PF00187 (chitin binding 1) (16). Analysis of the protein sequences with InterPro predicted a secretory signal peptide in both the Cccbm18-7-and Cccbm18-4-encoded proteins, indicating that they are trafficked outside the cells through the secretion pathway and may function at the cell surface. When fungal chitinase was added to the medium of C. cordycipiticola, the transcription levels of Cccbm18-4 and Cccbm18-7 increased significantly ( Fig. 6D and E). It was demonstrated previously that CBM18 genes can protect the fungus from chitinases in Batrachochytrium dendrobatidis (28). The chitinase-producing abilities of C. cordycipiticola and C. militaris were compared by staining with bromocresol purple, and it was found that C. militaris has a stronger ability to Cccbm18-7 (CCOR_09360) at CC and 4 and 8 dpi confirmed by qPCR. Gene expression was normalized to the expression of Cctef1 (CCOR_04235). Cccbm18-4 and Cccbm18-7 were upregulated at 4 dpi compared with CC. Error bars indicate standard deviations. (D and E) Transcript levels of Cccbm18-4 and Cccbm18-7 under chitinase stress determined by semiquantitative analysis. Gene expression was normalized to the expression of Cctef1. *** indicates a significant difference (P , 0.001). (F) Chitin degradation by C. cordycipiticola and C. militaris. C. militaris has a stronger ability to produce chitinase than C. cordycipiticola.
Interactions between C. cordycipiticola and Its Host Microbiology Spectrum produce chitinase than C. cordycipiticola (Fig. 6F). It was hypothesized that CBM18 proteins in C. cordycipiticola can bind chitin in its cell wall and thereby prevent chitinase hydrolysis by C. militaris.
Other than GH76 (a-mannanase) genes, some GH16 (b-glucanase) and GH18 (chitinase) genes were also important for the degradation of the cell wall of the host. Among the 16 GH16-encoding genes, 7 were upregulated. CCOR_10348, CCOR_05118, and CCOR_09367 were upregulated over 4-fold at 4 dpi compared with CC and maintained high expression levels at 8 dpi (Table S10). The content of glucan in the cell wall of C. militaris was 52 to 60% (26). The transcripts of the chitinase genes CCOR_07055, CCOR_10110, CCOR_08352, and CCOR_00575 were upregulated over 3-fold at 4 dpi compared with CC, and high expression levels were maintained at 8 dpi, except for the expression of CCOR_00575, which decreased slightly at 8 dpi compared to that at 4 dpi (Table S10).
A total of 330 CAZyme-related genes were predicted in the genome of C. militaris, among which 107 (107/330; 32.4%) were differentially expressed at 4 and 8 dpi compared with CM (Table S10). These genes were grouped into three main clusters according to the expression patterns (Fig. 6B). Ten genes were grouped into cluster I and showed upregulation at 4 and 8 dpi. There were 48 (44.86%; 48/107) genes in cluster III, and they were downregulated after infection. The genes of cluster II (45.80%; 49/ 107) were expressed at the highest levels at 8 dpi ( Fig. 6B; Table S10). The most upregulated CAZyme gene at 4 and 8 dpi was CCM_04420, which had very low expression levels at CM and was upregulated by over 4-and 6-fold at 4 and 8 dpi, respectively (Table S10). CCM_04420 was classified as a carbohydrate esterase family 10 (CE10) gene. This family is currently canceled in the CAZyme classification as members appear to act on ester-based modifications of compounds other than carbohydrates (29). Actually, CCM_04420 contained an acetyl esterase/lipase domain that was related to lipid transport and metabolism. Further analysis found that there were no homologs of this protein in the other species of Cordycipitaceae. The results of the phylogenetic analysis based on the protein sequences was totally incongruent with the species tree, and this gene was probably inherited from Diaporthomycetidae fungi (Fig. S4). Twenty-two GH18 genes were predicted in the genome of C. militaris (Table S11), more than in C. cordycipiticola. However, only 2 of them were downregulated at 4 dpi (Table S11), indicating that most chitinase genes did not respond to infection, although it was confirmed that C. militaris can produce chitinases at the mycelial stage (Fig. 6F).
Responses of genes related to SMs in Calcarisporium cordycipiticola and Cordyceps militaris. Among the 82 core genes of 66 SM clusters of C. cordycipiticola (16), 62 showed the lowest expression levels at 4 dpi (Table S12). It seemed that at 4 dpi, C. cordycipiticola grew mainly for colonization, and most of the SMs were inhibited temporarily. Sixteen core genes showed the lowest expression levels at CC and then showed an increasing trend at 4 or 8 dpi (Fig. S5 and Table S12), which should be related to interactions.
It was assumed that the CCOR_00730 (Ccnps6) cluster was responsible for the biosynthesis of siderophores in C. cordycipiticola (16). There were 14 genes in the Ccnps6 gene cluster (16), and most of them were highly expressed at the infection stage ( Fig. 7A; Table S13). It was inferred that C. cordycipiticola upregulated the transcripts of genes of the Ccnps6 cluster and synthesized a high level of siderophores during infection. A homologous cluster was also found in C. militaris, but it did not respond to infection (Table S13). The transcripts of genes of the Ccnps6 cluster in C. cordycipiticola were highly induced by iron deficiency (Fig. 7B). The pathogen was in an iron-deficient environment at the initial stage of infection, and the genes of the Ccnps6 cluster were induced for iron acquisition. The exogenous addition of a certain amount of iron (1, 5, and 10 mM FeSO 4 ) distinctly increased pathogenicity (Fig. 7C), indicating that iron was an important factor for the virulence of C. cordycipiticola.
Among the 40 core genes of 30 SM clusters of C. militaris, 6 showed the lowest expression levels at CM and then showed an increasing trend at 4 and 8 dpi (Fig. S5 and Table S12). These genes and their products should be related to interactions.
Among them, only 1 core gene (NRPS-like) (CCM_08331) showed .3-fold upregulation at both 4 and 8 dpi. Pfam annotated this product as "pyoverdine/dityrosine biosynthesis protein," which may be an efficient iron(III) transporter. More than one-half of the core genes in C. militaris showed deceased transcription with infection.
ROS-mediated interactions between Cordyceps militaris and Calcarisporium cordycipiticola. GO and KEGG analyses revealed that certain redox-related terms or pathways were enriched during infection in both C. militaris and C. cordycipiticola ( Fig. 4; Table S2). The concentration of hydrogen peroxide increased significantly in the C. militaris fruiting bodies after being infected by C. cordycipiticola and accumulated with the extension of infection (Fig. 8A), but no ROS accumulated in the hyphae of C. cordycipiticola as revealed by diaminobenzidine (DAB) and nitroblue tetrazolium chloride (NBT) staining (Fig. 8B). C. militaris can produce excessive ROS to respond to infection, which was similar to the immunity reaction of some plants responding to the pathogen. However, no obvious accumulation of ferric iron was observed in the infected tissues by staining with Prussian blue (Fig. 8A).
C. cordycipiticola is more sensitive to H 2 O 2 than C. militaris (Fig. S6), indicating that the pathogen C. cordycipiticola can remove H 2 O 2 from the environment by some mechanisms. Other than the upregulation of the Ccnps6 cluster and Cccfem genes ( Fig. 7A; Table S5), to acquire an appropriate amount of iron, genes related to antioxidant activity should make a contribution.
Eighty-six genes related to antioxidant activity (superoxide dismutase [SOD], catalase [CAT], glutathione, thioredoxin, and peroxidase systems) were identified in C. cordycipiticola ( Fig. 8C; Table S14), in which more peroxidases and thioredoxins were identified than in C. militaris. A total of 95.35% (82/86) of the genes were expressed  (Table S14), and 31 DEGs were grouped into three clusters (Fig. S7). Genes of clusters I and II were upregulated after infection and can remove ROS produced by the host during interactions. Sixty-one genes related to antioxidant activity were identified in C. militaris, and 90.16% (55/61) were expressed ( Fig. 8C; Table S14). Among these genes, 14 were differentially expressed during at least one stage (Table S14) and were grouped into 3 clusters, among which upregulated genes after infection were dominant (50%) (Fig. S7).

DISCUSSION
Calcarisporiaceae and Cordycipitaceae are sister families in Hypocreales (30). The parasite C. cordycipiticola and the host C. militaris are two phylogenetically close species. Here, time course dual RNA-seq was performed to reveal the pathogenic mechanisms of C. cordycipiticola and the responses of C. militaris to infection. C. cordycipiticola and C. militaris were in a dynamic battle, and the host C. militaris responded more slowly than the pathogen. C. cordycipiticola may secrete a large number of effectors to invade the host and evade the defense reactions of the host. Iron and ROS regulated the interaction between C. cordycipiticola and C. militaris. The Cmhsp70, Cmhsp78, and Cmhyd1 genes of C. militaris responded to the infection quickly, and they may be used as target genes for the breeding of resistant varieties. This study will be helpful for understanding the mycoparasitic interactions of two phylogenetically close species in their natural setting.
In our previous study, the process of infection of C. cordycipiticola in the fruiting bodies of C. militaris was monitored (16). It was found that the mycelia of C. cordycipiticola proliferated on the surface of the fruiting bodies of C. militaris at 3 to 4 dpi, and visible mycelia of C. cordycipiticola appeared on the fruiting bodies of C. militaris at 4 dpi. The color of the interior of the fruiting bodies gradually became deeper at 7 to 8 dpi. These two stages were key nodes of the infection. However, different infection stages should have their own controls in the best experimental scheme, which we will pay more attention to in our future research.
The pathogenic strategies of C. cordycipiticola may include (i) recognition and sensing of the host, (ii) colonization by rapid growth, (iii) suppression of the immune response of the host, (iv) nutrient absorption, (v) killing and degradation of the host, and (vi) adaptation to the microenvironment of the host. The first key step in a successful mycoparasite's life history is the ability to recognize and sense its host (31). Since early samples for transcriptome analysis were selected at 4 dpi, there was a possibility that the differential expression of genes responsible for recognition was not detected. As for the present data, lectins of pathogens may play a more important role in early recognition (see Table S9 in the supplemental material). The pathogen then regulated the expression of ribosome synthesis genes to synthesize a large number of proteins and promoted its growth and development. More secreted proteins were produced through the type II secretion system (Fig. 4), some of which continued to decompose the host cells and obtain nutrients and others of which surrounded their own cell walls to stabilize the internal environment for their growth.
In general, the host C. militaris responded more slowly than the pathogen (Fig. 3). Since C. cordycipiticola is the invader, it needs to mobilize more resources to invade and seize favorable ecological niches at the early stage of infection (4 dpi). Although the biomass of C. cordycipiticola was not high at the early stage, its gene expression was active, while C. militaris was in a defensive state at 4 dpi and was slow to respond to the outside world, and not many genes were mobilized. Later (8 dpi), the infection was severe, and more defense genes of C. militaris were mobilized to fight the pathogen.
During the colonization of the host, CAZymes were produced by both C. cordycipiticola and C. militaris (Table S10). However, C. cordycipiticola upregulated some CAZyme

Interactions between C. cordycipiticola and Its Host
Microbiology Spectrum genes quickly at 4 dpi and slowly in the host C. militaris, which was consistent with the transcription trends of most DEGs. C. cordycipiticola is a destructive mycoparasite that kills the host mainly by secreting cell wall-degrading enzymes (GH2, GH16, GH18, and GH76) and extracellular metalloproteases, etc. Finally, the cells of C. militaris were broken and died, leading to the wilting of fruiting bodies (16). The top 3 upregulated CAZyme genes at 4 dpi in C. cordycipiticola were CCOR_03676 (GH76), CCOR_09360, and CCOR_06535 (CBM18), whereas CCM_04420 (CE10), CCM_06273 (CE5), and CCM_08571 (GH76) were the top 3 upregulated CAZyme gene at 4 dpi in C. militaris. Since C. militaris has a stronger ability to produce chitinase than C. cordycipiticola (Fig. 6D to F), it was speculated that C. cordycipiticola can secrete CBM18 proteins with a chitin-binding domain to bind chitin on its cell wall and thereby prevent chitinase hydrolysis by the host C. militaris, which may be part of defending host immunity. This strategy has also been found for the plant pathogen Plasmodiophora brassicae (32). "Heme binding" (GO:0020037) was the only enriched GO term shared by both the pathogen and the host at both stages at 4 dpi and 8 dpi compared with CC or CM (Fig. 4). This GO term was defined as "binding to a heme, a compound composed of iron complexed in a porphyrin (tetrapyrrole) ring." Four pathways for iron acquisition have been identified in fungi: the siderophore-dependent pathway, the reductive pathway, the low-affinity ferrous iron uptake pathway, and the heme assimilation pathway (33). Heme constitutes a major source of iron for microorganisms (34), indicating that iron is important for both the pathogen C. cordycipiticola and the host C. militaris. Due to the scarcity of free iron in the host environment, pathogenic microorganisms by necessity have evolved mechanisms to extract iron from the host. Here, C. cordycipiticola was confirmed to upregulate the transcription of not only genes of the Ccnps6 cluster to produce many more siderophores to acquire iron from the host but also genes of the heme assimilation pathway. It was reported previously that CFEM proteins contributed to the acquisition of two forms of iron, heme and ferric ion, in the entire life cycle of Beauveria bassiana (33). The CFEM family members of Verticillium dahliae diverged to have different functions in iron predation and immunity suppression, both of which were necessary for colonization and propagation in the iron-insufficient environment of host plants (35). It was inferred that CFEM proteins may also help C. cordycipiticola with iron acquisition. There was an entirely different response of the host C. militaris since it can obtain iron from the medium. Genes of the siderophore cluster did not respond to infection. However, it may regulate the heme assimilation pathway to compete with the pathogen. This should be one of the defense mechanisms.
In plant-and animal-pathogenic fungi, iron has been characterized as a virulence determinant (36). The function of iron is still questioned in fungus-fungus interactions. In this study, we confirmed that iron is also a virulence determinant in C. cordycipiticola infecting C. militaris, which is the first report that iron regulates the virulence of pathogens on mushrooms. Increases in the iron concentrations in the air will accelerate outbreaks of this disease. It was concluded that avoiding the use of iron frames is beneficial for disease control during the cultivation of C. militaris.
During the interaction between C. militaris and C. cordycipiticola, ROS played an important role in both the defense response of C. militaris and infection by the pathogen. When C. cordycipiticola invades, the host C. militaris may, like plant hosts such as rice, induce the expression of its disease resistance genes and produce a hypersensitive response through an oxygen burst, thereby preventing the further expansion of the pathogen. Invasion by C. cordycipiticola resulted in the accumulation of excessive ROS in the hyphae of fruiting bodies of C. militaris (Fig. 8B), causing much molecular damage such as protein oxidation, DNA mutation, and lipid peroxidation. During the infection process of C. cordycipiticola, a small number of hyphae will penetrate the hyphae of C. militaris (16), exposing it to an environment with a high level of reactive oxygen, but the hyphae can still grow normally (16), indicating that the pathogen should be able to scavenge reactive oxygen and adapt to the host environment. Genes related to ROS scavenging in C. cordycipiticola were also differentially expressed at different levels, and these genes may play an important role in regulating the homeostasis of ROS in their hyphae.
PAMPs can efficiently eliminate invading pathogens and have been researched in detail in both plant and animal hosts. Here, there was no distinct differential expression of 18 lectin-encoding genes at both 4 and 8 dpi compared to CM for the host (Table  S9). Whether the host C. militaris could perceive the presence of pathogens well is worth studying in the future since there are almost no reports on the PAMPs of mushrooms infected by fungal pathogens.
The pathogen's attack substantially altered the primary metabolism of the host C. militaris, as seen by the fact that only KEGG pathways of amino acid and 2-oxocarboxylic acid (Table S2) were enriched at both 4 and 8 dpi. It was also reported previously that Rhizoctonia infection alters the primary metabolism of soybean seedlings, including carbohydrate and amino and carboxylic acid metabolisms (37).
The primary and secondary metabolisms of both the pathogen C. cordycipiticola and the host C. militaris were changed during the interaction process. The difference is that the pathogen enhances them to promote colonization, while primary and secondary metabolic disorders occurred in the host C. militaris. Other than producing excess ROS, the defense mechanisms of C. militaris may include the quick upregulation of Cmhsp78, Cmhsp70, and Cmhyd1. In addition to serving as molecular chaperones, HSPs have been found to play essential and regulatory roles in the innate immune responses of plant (38) and animal (39) cells. According to our previous study on the function of Cmhyd1 (21), it was speculated that CmHYD1 may be secreted to the mycelial surface of C. militaris, hinder the recognition of the pathogen, and, thus, improve the disease resistance of C. militaris. No C. militaris strain resistant to C. cordycipiticola has been found; there is an urgent need to explore potential resistance genes for the breeding of C. militaris. Cmhsp78, Cmhsp70, and Cmhyd1 may be candidate genes.

MATERIALS AND METHODS
Strains and culture conditions. C. cordycipiticola strain CGMCC 5.2193 was isolated from diseased fruiting bodies of C. militaris strain CGMCC 3.16320 in our laboratory (16). Both strains were maintained on potato dextrose agar (PDA) at 20°C. The fruiting bodies of C. militaris were cultivated according to previously reported methods (40).
Collection of samples for dual RNA-seq. C. cordycipiticola was grown on PDA medium in the dark at 20°C for 7 days and then used to infect 4 to 5-cm-long fruiting bodies of C. militaris. Infection was performed according to our previously described methods (16), with modifications. Discs of mycelia (1 cm in diameter) of C. cordycipiticola cultured on PDA were cut with a puncher. Ten pieces of mycelia were inoculated onto the surface of fruiting bodies in each cultivation bottle. C. militaris pathogenic sites (1 cm 2 ) were harvested after infection for 4 and 8 days and named as 4 dpi and 8 dpi, respectively. The mycelia of C. cordycipiticola grown on PDA medium covered with a sheet of sterile cellophane in the dark at 20°C for 7 days (CC) and the fruiting bodies of C. militaris (CM) used for infection were harvested as controls. CC was collected by scratching fungal mycelia from the cellophane using a sterile spatula. CM was harvested and cut into small pieces. All samples were immediately frozen in liquid nitrogen and stored at 280°C until RNA was extracted. Samples from each of the three different bottles were defined as three biological replicates for each treatment. Three biological replicates were included for each treatment, and 12 samples were used for dual-RNA-seq analysis.
RNA isolation, library construction, and sequencing. Dual RNA-seq was performed according to the flow diagram shown in Fig. S1 in the supplemental material. First, total RNA was extracted and sequenced. Next, clean reads were mapped to the reference genome sequences of C. cordycipiticola and C. militaris. Finally, gene transcription was analyzed as usual.
Total RNA was extracted using an Omega plant kit (Omega Engineering, Norwalk, CT, USA). The RNA concentration was measured using a NanoDrop 2000 instrument (Thermo Fisher Scientific, Waltham, MA, USA), and the RNA integrity was assessed using the RNA Nano 6000 assay kit for the Agilent Bioanalyzer 2100 system (Agilent Technologies Inc., Palo Alto, CA, USA). Twelve libraries were generated using a NEB Next Ultra RNA library prep kit (New England BioLabs [Beijing] Ltd., Beijing, China) according to the manufacturer's recommendations and sequenced (paired end, 150 bp each) on an Illumina HiSeqX-ten platform (Illumina Inc., San Diego, CA, USA) by Biomarker Technology Co. Ltd. (Beijing, China).
Read mapping and RNA-seq statistical analysis. Raw data (raw reads) in fastq format were first processed through in-house Perl scripts based on FastQC (https://raw.githubusercontent.com/s-andrews/FastQC/ master/INSTALL.txt). In this step, clean data (clean reads) were obtained by removing reads containing the adapter or poly(N) and low-quality reads from the raw data. All of the downstream analyses were based on clean data of high quality.
Interactions between C. cordycipiticola and Its Host Microbiology Spectrum For the pathogen C. cordycipiticola and the host C. militaris, clean reads were aligned to the reference genomes C. cordycipiticola CC01 (16) and C. militaris CM01 (41), respectively, using HISAT2 (42). The above-described reads were assembled using StringTie (43). The gene expression levels were estimated by the FPKM.
Pearson's correlation coefficients of gene expression between replicates were used to determine whether the biological replicates were sufficiently similar for subsequent statistical analysis (44). The host and pathogen transcriptomes were subjected to differential expression analysis using the DESeq2 R package (45). A corrected false discovery rate (FDR) of 0.01 and a log 2 fold change of 1 were set as the thresholds for differential expression unless specified otherwise.
Functional annotation of DEGs. The GO enrichment analysis of the DEGs was performed using the GOseq R package based on a Wallenius noncentral hypergeometric distribution, which can be adjusted for gene length bias in DEGs (46). The redundancy in the significantly enriched GO terms was eliminated by using REVIGO (http://revigo.irb.hr/). KOBAS software was used to test the statistically significant enrichment of DEGs in KEGG pathways (47).
Reverse transcription-PCR and qPCR analyses. The synthesis of cDNA from RNA (1 mg) was carried out using a HiScript III 1st-strand cDNA synthesis kit (1gDNA wiper) (Vazyme Biotech Co. Ltd., Beijing, China). For the semiquantitative analysis, the intensity of each band in the gel images was measured using ImageJ software (https://imagej.net/ij/index.html), and the ratio of the expression level of the tested gene to the expression level of Cctef1 (CCOR_04235) was calculated. The semiquantitative data sets with and without chitinase were compared via Student's t test.
qPCR was conducted using AceQ qPCR SYBR green master mix (Q111) (Vazyme Biotech Co. Ltd., Beijing, China) with the CFX Connect real-time system (Bio-Rad, Hercules, CA, USA). The relative gene expression levels were calculated using the 2 2DDCT method (48). The Cctef1 (CCOR_04235) and Cmrpb1 (CCM_05485) genes were used as internal standards for C. cordycipiticola and C. militaris, respectively. All of the primer sequences used in this study are listed in Table S7. The obtained data represent the results from three biological replicates, with two technical replicates each.
Determination of chitinase activity. Chitinase activity was determined using bromocresol purple staining according to methods described previously (49), with modifications. Conidia of C. cordycipiticola (1 Â 10 6 ) were inoculated into medium with chitin colloid as the sole carbon source and bromocresol purple as an indicator and then cultured at 20°C. The color of the tested plates was observed.
ROS detection. The localization of ROS in C. militaris after infection for 4 and 8 days was visualized using DAB and NBT stainings, which were performed as described previously, with slight modifications (50). Briefly, sections with a thickness of 5 to 10 mm were cut transversely under a microtome (EM UC7; Leica, Wetzlar, Germany), treated with 0.05% (wt/vol) DAB and NBT in 50 mM phosphate-buffered saline (pH 7.4), and incubated for 20 min at 20°C. The buffer was then removed, and the sections with C. cordycipiticola were washed three times with distilled water. The slices were placed onto absorbent blot paper, observed under an anatomical lens and a light microscope, and photographed.
Virulence assays and exogenous iron addition. For the evaluation of the virulence of C. cordycipiticola, C. militaris fruiting bodies of 3 to 4 cm in length were spray inoculated with 2 mL of a spore suspension (5 Â 10 6 spores/mL) of C. cordycipiticola in 0.05% Tween 80. Inoculated C. militaris fruiting bodies were kept in a growth chamber under 12 h of light/12 h of darkness at 20°C with 90% relative humidity. Virulence was evaluated 15 days after inoculation, and the infected fruiting bodies were harvested from three independent tissue culture vessels. Ten infected fruiting bodies were randomly chosen from each tissue culture vessel, and the area of each lesion was measured by using Nano Measurer 1.2 (51).
To determine the effect of iron on infection by C. cordycipiticola, 3 mL of an iron solution (0, 1, 5, and 10 mM FeSO 4 solution in 0.05% Tween 80) and C. cordycipiticola conidia were sprayed onto the surface of C. militaris fruiting bodies. The fruiting bodies sprayed with FeSO 4 (0, 1, 5, and 10 mM FeSO 4 solution in 0.05% Tween 80) were used as controls. Virulence assays were performed as described above.
Data availability. The raw data were deposited in GenBank under BioProject accession number PRJNA930069.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 6.7 MB.

ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (grant number 31872163), the Inner Mongolia Autonomous Region Science and Technology Project (project identifier 2022YFDZ0092), and the CAS Engineering Laboratory for Advanced Microbial Technology of Agriculture (project identifier KFJ-PTXM-016).
We declare no competing interest.