The differences in fungal spore concentration were adjusted to ensure uniform and comparable development of lesions in tomato fruit. In the case of R. stolonifer inoculations of MG fruit, we also tested a concentration of 1,000 spores/µl but no differences in fruit responses or fungal growth between this concentration and 30 spores/µl were observed. Inoculated fruit were incubated at RT in high humidity chambers. For mock inoculations, the same procedure was followed but without the addition of the inoculum. The pericarp and epidermis of the blossom end were collected at 1 and 3 days post-inoculation , immediately frozen in liquid nitrogen, and stored at −80◦C until use. One biological replicate consisted on average of eight fruit, and five biological replicates per treatment were obtained.Tomato fruit tissues were ground using a Retsch R Mixer Mill MM 400 and RNA was extracted from 1 g of fine-powdered tissue according to the procedure described in Blanco-Ulate et al. . Fungal RNA from the in vitro cultures was extracted using TRIzol and purified using the Quick-RNA MiniPrep Kit following the procedure described in MoralesCruz et al. . The RNA concentration and purity were assessed with the Qubit 3 and the NanoDrop One Spectrophotometer , respectively. Gel electrophoresis was used to confirm the RNA was not degraded. Barcoded cDNA libraries were prepared using the Illumina TruSeq RNA Sample Preparation Kit v2 . Quality control of the cDNA libraries was performed with the High Sensitivity DNA Analysis Kit in the Agilent 2100 Bioanalyzer . 50-bp single-end libraries were sequenced on the Illumina HiSeq 4000 platform in the DNA Technologies Core of the UC Davis Genome Center. In total, 18 libraries were sequenced for B. cinerea , 17 libraries were sequenced for F. acuminatum , and 17 libraries for R. stolonifer .

Quality trimming of the raw reads was performed with sickle v1.331 and adapter sequences were removed with scythe v0.9912 . The basic quality measurements were assessed with FastQC3 . To generate transcriptome assemblies for F. acuminatum and R. stolonifer, stacking flower pot tower reads from samples infected with each of these pathogens were mapped to the tomato genome using STAR 2.6 . Reads that failed to map to the tomato gene were pooled with the reads from the in vitro samples and used for de novo transcriptome assembly. Transcriptomes were assembled with Trinity 2.5.1 using default parameters . Quality of the assemblies was assessed with BUSCO 3.0.2 using the “fungi odb9” dataset, while basic assembly metrics were obtained with Transrate 1.0.3 . Potential contaminanttranscripts were identified via BLAST using both the blast nr database and the UniProt database. Functional annotations for transcriptomes of all three fungi were created using Trinotate 3.0.1 . The default Trinotate parameters were used to retrieve the best BLAST hits as well as annotations for Gene Ontology terms, Pfam families, Kyoto Encyclopedia of Genes and Genome pathways, EggNOG predictions, and SignalP sequences. Custom BLAST databases were incorporated according to the Trinotate manual for the Transporter Classification Database and the Pathogen-Host Interactions database . Custom HMMER alignment results for HMM profiles from dbCAN7 and fPoxDB8 were incorporated similarly.To determine if F. acuminatum and R. stolonifer show similar patterns of infections in tomato fruit as B. cinerea , we did side-by-side inoculations of fruit at two developmental stages: unripe and ripe . As displayed in Figure 1A, we confirmed that all fungi were unable to cause rotting in MG fruit but aggressively colonized RR fruit. These results were further validated by quantifying fungal biomass based on relative expression of fungal reference genes via qRT-PCR .

At 3 dpi, RR fruit inoculated with B. cinerea and F. acuminatum showed water-soaked lesions of approximately 15 mm covered by dense mycelia, whereas RR fruit inoculated with R. stolonifer were almost decomposed and entirely covered by mycelia. Although no lesions were observed in MG fruit when inoculated with any of the pathogens, some differences in fungal growth and tomato responses were observed. Inoculations with B. cinerea and R. stolonifer did not show any visible mycelia, whereas F. acuminatum inoculations showed limited hyphal growth without disease symptoms. All three fungi induced a necrotic ring surrounding the inoculation sites during the incompatible interaction with MG fruit, yet F. acuminatum inoculations caused dark and wide rings while fruit infected with R. stolonifer developed a weaker response. Because we were not able to visually detect any hyphal growth of B. cinerea and R. stolonifer inMG fruit, we used a microscope to observe whether the spores germinated in the inoculated wounds. At 1 dpi, B. cinerea spores were mainly ungerminated or in the process of germination . By contrast, F. acuminatum and R. stolonifer already showed active hyphal growth, indicating that spores of these fungi germinate earlier on MG fruit. At 3 dpi, some hyphal growth was also observed for B. cinerea. Together, these observations suggest that the incompatibility of the interaction between these fungi and MG tomato fruit occurs after spore germination. To provide initial support that both F. acuminatum and R. stolonifer are capable of inducing disease responses in the host, like B. cinerea, and do not merely behave as saprotrophs , we evaluated the expression of the host gene SlWRKY33 , which is well-known to be pathogen-responsive but is not induced by abiotic stresses . To test that the induction of this gene occurred only as a result of inoculation and not wounding, we included a mock-inoculated control in our analyses. The expression patterns of SlWRKY33 measured by qRT-PCR reflected the accumulation of fungal biomass and the presence of lesions in each of the treatments . At 1 dpi, expression of SlWRKY33 was induced by inoculation with both B. cinerea and F. acuminatum but not with R. stolonifer or mock inoculation in MG fruit. In RR fruit, pathogen-induced SlWRKY33 was detected for all three pathogens at greater levels than found in MG fruit.Our observations of lesion development, fungal biomass, and activation of pathogen responses led to the hypothesis that F. acuminatum and R. stolonifer display a similar necrotrophic behavior in tomato fruit as B. cinerea. Therefore, to discover pathogenicity or virulence factors in these fungi that are important for necrotrophic infections, we performed a genomewide transcriptomic analysis of inoculated fruit at both time points as well as in vitro cultures. Due to the lack of publicly available genomic data for F. acuminatum and R. stolonifer, we assembled de novo transcriptomes for both of these pathogens from our cDNA libraries following the Trinity pipeline . Using the fungal ortholog dataset of the Benchmarking Universal Single-Copy Orthologs tool , we determined that our assemblies presented high completeness, with 88.2 and 90.3% of F. acuminatum and R. stolonifer matches being complete, respectively. Our F. acuminatum transcriptome contained 20,117 unique transcripts, danish trolley while our R. stolonifer transcriptome contained 19,754 . We then used homology-based annotation to obtain information on gene functions for each of the transcriptomes, including the B. cinerea B05.10 ASM83294v1 . We annotated transcripts based on nine separate functional classifications, including GO , Pfam domains , Pathogen-Host Interaction , membrane transporters , Carbohydrate-Active Enzymes , and fungal peroxidases .

Each type of functional annotation was represented by a similar percentage of annotated transcripts across all pathogens . The specialized enzyme classifications of peroxidases and CAZymes made up a relatively small fraction of the annotated transcripts, whereas general functional classifications such as GO, Pfam, and KEGG descriptions were available for at least 70% of the annotated transcripts for all pathogens. Annotations for all three transcriptomes can be found in Supplementary Table S3. Although the F. acuminatum and R. stolonifer transcriptomes are preliminary and may require further curation and validation, we consider that they are a valuable resource to perform gene expression analyses and to shed light on the infection strategies utilized by these fungi.First, we performed principal component analysis to determine if the fungal-inoculated and in vitro samples could be discerned based on the expression of the fungal transcripts. The PCAs revealed that all samples clustered by treatment . In most cases, the first component clearly differentiated the MG fruit from the RR fruit inoculations and the in vitro samples. Then, we determined DEGs between inoculations of MG or RR fruit and in vitro cultures for each pathogen. Across all comparisons, we detected 6,488 B. cinerea DEGs , 6,154 F. acuminatum DEGs , and 8,777 R. stolonifer DEGs . The number of DEGs for R. stolonifer were mainly identified in the RR fruit comparisons, as the low amount of fungal biomass in MG fruit samples did not allow for an in-depth sequencing coverage of the fungal transcripts. To confirm the accuracy of the DEG analysis, we selected a subset of genes for each pathogen to validate their expression using a qRT-PCR approach . Our results confirmed that the gene expression values were consistent, showing significant Pearson correlation coefficients and between the RNA-seq and the qPCR expression data . We further evaluated the fungal DEGs based on whether they were commonly or uniquely expressed under specific treatments, which can provide insight on particular sets of genes that are relevant during incompatible or compatible interactions . For each pathogen, genes uniquely upregulated in RR fruit constituted a sizable fraction of upregulated genes . This result may be influenced by the fact that RR fruit samples had more coverage of fungal transcripts in the RNA-seq experiment than MG fruit samples, which is a technical limitation of this type of study. Nevertheless, the comparisons of common and unique DEGs among treatments for each of the pathogens support the results of the PCAs, indicating that these fungi display a specific behavior in each of the fruit stages at early and late time points after inoculation. We also identified upregulated DEGs shared across categories that are likely to represent core pathogenicity factors during fruit infections.To gain insight into key biological processes that are relevant during compatible or incompatible fruit infections, we performed GO enrichment analyses of the upregulated DEGs in all combinations of ripening stage and dpi for each pathogen . We mainly focused on GO terms of the “biological process” class that were significantly enriched and appeared to be involved in pathogenesis or fungal growth in the host tissues . Upregulated DEGs from all comparisons, except for R. stolonifer MG inoculations, were enriched in oxidation reduction processes . A closer inspection of these DEGs revealed functions that are likely to be involved with pathogenicity, such as catabolism of ROS [e.g., superoxide dismutases , catalases , peroxidases] and breakdown of cell wall molecules such as cellobiose and lignin . In B. cinerea, the SOD BcSOD1 was induced in both MG and RR fruit at 1 and 3 dpi. Additionally, BcSOD3 was upregulated only in MG fruit at 1 dpi, and BcSOD2 is upregulated only at 1 dpi in MG and RR fruit. Although two potential SODs, FacuDN9613c0g1i1 and FacuDN4275c0g1i2, were employed by F. acuminatum in all treatments except 1 dpi MG, none of the seven putative SODs identified in R. stolonifer were upregulated in any of the treatments. To further identify enzymatic scavengers of hydrogen peroxide , we examined the upregulated DEGs of each pathogen which showed significant similarity to members of the Fungal Peroxidase Database. This analysis revealed differences both in the classes of enzymes used in each pathogen and the treatments in which they were used. For example, in B. cinerea, only two known catalases, BcCAT2 and BcCAT4 , were found to be upregulated during tomato fruit interaction. Both of these were only active in MG fruit. In contrast, F. acuminatum exhibited very strong induction of two predicted CATs, FacuDN12367c0g1i1 and FacuDN13048c0g1i1, at 1 dpi in RR fruit but not in MG fruit, although a handful of CATs and catalase-peroxidases were upregulated less strongly across both MG and RR fruit. In all F. acuminatum-inoculated samples, there was also an enrichment of DEGs involved in hydrogen peroxide catabolism , further highlighting the importance of fungal responses to oxidative stress during fruit colonization. In R. stolonifer, peroxidases were only upregulated at 1 dpi in RR fruit and included two 2-cysteine peroxiredoxins , one cytochrome C peroxidase, and one glutathione peroxidase . Additionally, in all B. cinerea-inoculated samples, DEGs annotated with the oxidation-reduction process GO term included enzymes in the biosynthetic pathways for the phytotoxins botrydial and botcinic acid.