A novel Arabidopsis pathosystem reveals cooperation of multiple hormonal response-pathways in host resistance against the global crop destroyer Macrophomina phaseolina

Dubbed as a “global destroyer of crops”, the soil-borne fungus Macrophomina phaseolina (Mp) infects more than 500 plant species including many economically important cash crops. Host defenses against infection by this pathogen are poorly understood. We established interactions between Mp and Arabidopsis thaliana (Arabidopsis) as a model system to quantitatively assess host factors affecting the outcome of Mp infections. Using agar plate-based infection assays with different Arabidopsis genotypes, we found signaling mechanisms dependent on the plant hormones ethylene, jasmonic acid and salicylic acid to control host defense against this pathogen. By profiling host transcripts in Mp-infected roots of the wild-type Arabidopsis accession Col-0 and ein2/jar1, an ethylene/jasmonic acid-signaling deficient mutant that exhibits enhanced susceptibility to this pathogen, we identified hundreds of genes potentially contributing to a diverse array of defense responses, which seem coordinated by complex interplay between multiple hormonal response-pathways. Our results establish Mp/Arabidopsis interactions as a useful model pathosystem, allowing for application of the vast genomics-related resources of this versatile model plant to the systematic investigation of previously understudied host defenses against a major crop plant pathogen.


A quantitative assay system for Mp-infected Arabidopsis roots. A fundamental step in the infection of plants by
Mp is the establishment of hyphae and microsclerotia in plant tissue. We observed this process in Mp/Arabidopsis interactions on agar plates. Mp microsclerotia added to ½ MS agar generate hyphae that grow throughout the agar, forming more microsclerotia as Mp grows. When Arabidopsis seedlings are laid down upon the Mp-laden agar infection plate, hyphae on the agar surface contact the roots and grow toward them, surrounding and penetrating the root tissue (Fig. 1A,D). Microsclerotia can be seen forming within the roots as early as 48 hours post contact (hpc) in Arabidopsis (Fig. 1E).
The fact that critical events of the Mp infection process occur in Arabidopsis roots under controlled conditions on agar enables the design of plate-based infection protocols to accurately quantify levels of host susceptibility ( Supplementary Fig. S1). Here, Arabidopsis seeds were sown on ½ MS agar plates ( Supplementary  Fig. S1A) and allowed to grow for ten days ( Supplementary Fig. S1C) while Mp microsclerotia were added to separate plates ( Supplementary Fig. S1B) and grown for six days (Supplementary Fig. S1D). The ten-day-old seedlings were transferred to the six-day-old Mp infection plates ( Supplementary Fig. S1E). After 24 hours of contact ( Supplementary Fig. S1F), the seedling roots were infected by Mp hyphae (Fig. 1A,C). By 48 hours, hyphae penetration continued and microsclerotia could be seen forming in some Arabidopsis roots (Fig. 1E). Root tissue was flash frozen in liquid nitrogen at 24 hpc and 48 hpc for RNA-seq or at 48 hpc for qPCR analysis ( Supplementary  Fig. S1H). Alternatively, at four-five days post contact (dpc) microsclerotia density was determined by counting the number of microsclerotia per mm in each seedling's primary root ( Supplementary Fig. S1G). Infection plates typically contained 12-14 seedlings, and the microsclerotia per mm density was based on the average of all tested seedlings per plant line (at least three plates per line) in a biological replicate.

Arabidopsis mutants compromised in et signaling show enhanced root susceptibility to Mp.
To identify host defense mechanisms that affect levels of Mp susceptibility in Arabidopsis roots, we tested a set of Arabidopsis Col-0 mutants. As Mp is known to mainly exhibit a necrotrophic lifestyle, we tested the jar1 and ein2 single mutants as well as the ein2/jar1 double mutant with defects in signaling processes mediated by JA or/ and ET, respectively ( Fig. 2A). Mutants of JAR1 (Jasmonoyl isoleucine conjugate synthase1) are compromised in the conversion of (+)-7-iso-JA to (+)-7-iso-jasmonoyl-L-isoleucine (JA-Ile), which is one of the main bioactive forms of JA 47 . The jar1 mutant is known to exhibit complete loss of JA signaling in Arabidopsis roots 48 . The ER-associated EIN2 (ET INsensitive 2) protein links signaling processes triggered by various ET receptors to EIN3-EIL1-type transcription factors 49 . Null mutants of EIN2, including the ein2 lines used here, are completely ET insensitive 50 .
In the ein2/jar1 double mutant, but not in the ein2 or jar1 single mutants, we observed significantly increased density of microsclerotia in roots ( Fig. 2A). We confirmed this finding by comparing the abundance of Arabidopsis-and Mp-specific genomic DNA in Arabidopsis roots 48 h after Mp contact using quantitative PCR (qPCR) with primer pairs specifically targeting species-specific sequence characterized amplified regions (SCAR) of Mp (MpSyk) and Arabidopsis (Shaggy-related Kinase 11, AtSK11) DNA 51,52 (Fig. 2B). This SCAR-qPCR assay showed relative abundance of Mp-specific DNA (compared to Arabidopsis-specific DNA) to be clearly elevated in Mp-infected roots of the ein2/jar1 line compared to Col-0 plants indicating twice the amount of Mp biomass in this mutant. Contrary to our microsclerotia density assay, we also observed a significant increase of relative Mp DNA levels in the ein2 single mutant compared to Col-0 (Fig. 2B). We attribute this to the fact that the SCAR-qPCR assay is more sensitive than microsclerotia density measurements and detect DNA from other Mp structures, such as hyphae. While neither the microsclerotia density assay ( Fig. 2A), nor SCAR-qPCR (Fig. 2B) indicated significantly enhanced susceptibility in the jar1 single mutant, this mutant exhibited a large degree of www.nature.com/scientificreports www.nature.com/scientificreports/ Mp biomass variability in the SCAR-qPCR. We also included, in both assays, the transgenic NahG line 53 which does not accumulate the defense hormone SA, but did not observe any significant effect. However, as in the case of jar1, our SCAR-qPCR assay revealed that Mp biomass variability was unusually strong in NahG plants (Fig. 2B).
Collectively, in our agar plate-based assay system, we observed enhanced levels of Mp growth in mutants compromised in ET signaling, showing that this stress signaling pathway contributes to defense reactions in Arabidopsis roots against Mp.
Arabidopsis mutants compromised in et-, JA-and SA signaling show enhanced susceptibility to Mp in shoots. In order to investigate Mp-induced damage to aerial plant tissues, since shoot damage is a common visual indicator of disease in many Mp-infected crop plants in the field 16 , we repositioned the seedlings and increased the length of time on Mp infection plates. Despite the still-abundant level of Mp on infection plates, these few changes allowed for a more gradual development of disease symptoms with visible phenotype variation in shoots (i.e., differing rates of decay) between lines (Fig. 3). Similar to the root assays, ten day old Arabidopsis seedlings were transferred to Mp infection plates; however, plants were allowed to grow for two weeks with plates positioned right-side-up under growth room conditions. Seedlings were photographed during the infection period and their disease severity was determined. Using a disease index scoring system of "0", for healthy plants, through "5", for dead plants, (Fig. 3B), we compared disease symptoms (e.g., chlorosis, necrosis, stunting) between Col-0 and mutant lines. Of the six mutants with defects in ET and/or JA-signaling that we tested, five (ein2/sid2, ein2, ein3/eil1, ein2/jar1, jar1) exhibited more severe disease symptoms (average disease index scores of 4.5, 4.3, 4.0, 4.6, 3.3, respectively) than Col-0 (wild type, average disease index score: 2.26) after two weeks (Fig. 3C). In some cases, the effect was already clear after 1 week. After two weeks of Mp infection, ein2/jar1 had the most severe disease symptoms and many of the ein2/jar1 (81%), ein2 (70%) and jar1 (34%) plants were dying (disease index score 4.5-5; Col-0, 7%) (Fig. 3A). In addition to these mutants, which are compromised in upstream ET-and/or JA-signaling processes, we also observed clearly enhanced Mp susceptibility in the ein3/eil1 double mutant (Fig. 3C), which is deficient in EIN3 and EIL1 54 . Both of these related transcription factors control almost all ET-responses as well as some responses to JA 55 . However, we did not observe any significant change of Mp susceptibility levels in jin1, a mutant deficient in the transcription factor Myc2, which mediates JA-triggered wound responses 56 .
Shoot disease symptoms were less severe in SA-deficient NahG plants than in those of the JA and ET mutants, yet were also significantly higher than Col-0 plants at the two-week time point (Fig. 3C). To confirm a potential role of SA in host defenses against Mp we tested additional SA signaling deficient Arabidopsis lines and observed enhanced levels of Mp susceptibility relative to Col-0 in the sid2-2, wrky70 and wrky70/54 mutants (Fig. 3C). The sid2-2 mutant is deficient in isochorismate synthase 1, an enzyme catalyzing a key step in defense-induced SA biosynthesis 57 , while the tested wrky mutants are blocked in regulatory steps downstream from SA, with WRKY70 and WRKY54 being two closely related TFs mediating SA-responsive gene expression 58,59 . Given that SA contributes to elevated Mp tolerance in our experiments, the lack of significant effects of npr1 and pad4 plants  51 and Arabidopsis (Shaggyrelated Kinase 11, AtSK11) 52 by qPCR using Arabidopsis roots 48 hpc with Mp. Error bars represent standard error for three biological replicates. Statistical significance relative to Col-0 determined by Student's t-test, asterisk; p < 0.05.
in the shoot disease index assay was unanticipated. Yet, while NPR1 and PAD4 are both known to contribute to SA-dependent immunity, their significance for certain defense mechanisms seems variable. NPR1 is indispensable for SA-dependent systemic immunity, but its importance for local defense reactions may be limited 60 . The . Arabidopsis shoot disease symptom severity was higher in mutants than in Col-0. (A) Ten day-old seedlings were placed on ½ MS agar plates (left) or Mp infection plates (right) and scored 7 and 14 days postcontact (dpc) for shoot disease symptom severity. Representative plates (one untreated and one infection plate) are shown. (B) Arabidopsis shoot disease index system with representative plant score photos. (C) Plants were scored at 7 dpc (grey bars) and at 14 dpc (black bars) with Mp. Error bars represent standard error for at least three biological replicates, n = 30. Statistical significance relative to Col-0 was determined using the Wilcoxon Rank Sum test for two independent samples, asterisk; p < 0.05.

Profiling of Mp-induced transcriptome changes in Arabidopsis roots.
In order to identify host transcriptome changes associated with defense responses against Mp in Arabidopsis roots, we performed mRNA-seq with poly-A mRNA from agar plate-grown Arabidopsis Col-0 or ein2/jar1 roots at 24 h or 48 h after Mp contact or control treatment. For each of the eight different experimental conditions two independent biological replicates were performed. A comparison between levels of Arabidopsis and Mp-specific 18S RNAs at 48 h post contact in samples used for this analysis indicated a nearly two-fold higher abundance of Mp in ein2/jar1 roots compared to roots of Col-0 ( Supplementary Fig. S3). This observation further confirms that the ein2/jar1 double mutant exhibits elevated susceptibility to Mp and shows that the treatments for our transcript profiling experiment were successful.
To identify gene expression changes potentially relevant for the induction of heightened resistance in Col-0 we performed the following pair-wise comparisons: (1)  (Supplementary Tables S1-S12 and Fig. S4). Transcript level differences in these comparisons with an adjusted p value of less than 0.05 (Table 1) were considered significant.
The set of Arabidopsis Col-0 genes up-regulated in response to Mp infection at either 24 hpc or 48 hpc (3,396 genes) seems to a certain extent unique, as it shows only partial overlap with gene sets up-regulated by other pathogens, such as the necrotrophic fungi Fusarium oxysporum and Botrytis cinerea or the biotrophic oomycete Hyaloperonospora arbidopsidis (Hpa) (Fig. 4A). Of the three comparisons in Fig. 4A, the gene set induced by F. oxysporum 64 , has the largest relative overlap with the Mp-induced transcript profile in Col-0 roots, as 68% (79/116) of F. oxysporum-induced genes are also up-regulated by Mp. F. oxysporum and Mp are both necrotrophic, soil-borne fungal pathogens that often infect plants through their roots. The overlap of genes up-regulated in aerial parts of Arabidopsis by the nectrotrophic fungus B. cinerea 65 with Mp-induced genes is less extensive at 42% (687/1,637), which, however, may partially reflect the fact that different types of host tissues were used. A relatively low percentage (28%; 1,123/3,950) of Arabidopsis genes induced by Hpa 66 are also up-regulated by Mp. This is not surprising given that Hpa is a biotrophic oomycete that infects aerial portions of the plants and exhibits a very different lifestyle from Mp. Again, these differences also likely reflect that different host tissue types were profiled. Despite the fact that the relative overlap of these two gene sets is small, a large absolute number of genes (1,123) are jointly up-regulated in both types of infections indicating that some host defense responses may be inducible by both of these dissimilar types of pathogens.
Similarly, a comparison of the same set of Mp-induced genes to those up-regulated in hormone-treated Arabidopsis revealed commonalities of Mp-induced responses to those activated by other defense stimuli (  www.nature.com/scientificreports www.nature.com/scientificreports/ pathogen. TFs typically bind specifically to short DNA stretches of defined sequence, but also exhibit some ambiguity binding to a range of derivatives of a certain type of DNA motif. While any sequence motif is randomly distributed in genomic DNA, statistical enrichment of a given motif in promoter regions of genes that are co-expressed under certain conditions, can indicate a role for this motif (and cognate binding TFs) in directing this expression pattern 70 . JA-responses are often mediated by Myc2 and related bHLH TFs, which have a strong binding preference for the well-characterized G box (CACGTG) 56,71 . Transcriptional up-regulation in response to ET and JA is mediated by the related Arabidopsis TFs EIN3 and EIL1 54 . EIN3 (and likely also EIL1) can bind to DNA sequences containing the pentameric motif TACAT 72 . In response to ET and/or JA, EIN3/EIL1-related TFs can activate expression of members of the large ERF TF family which bind to a variety of binding sites related to the GCC box (AGCCGCC) 73 . WRKY TFs, which bind to various sequences related to the W box (TTGACC/T) 58,59 , are associated with SA-driven transcriptome changes, but can also be involved in other types of immune responsive gene expression control, such as responses mediated by PTI-associated MAP-kinase cascades. More strictly associated with SA-signaling seem to be TGA-bZIP TFs and TBF1 74,75 . While the former type of TFs are known to bind to TGA boxes (TGACG), TBF1 (and likely additional so far unknown factors) binds to TC-rich sequences, such as the TL1 element (TTCTTCTTC) 75,76 or the related LURP A element (TCTTCT) 69 .
Our set of RNA-seq data allowed us to discriminate three temporal patterns of Mp-induced transcript accumulation in Col-0 plants: (1) "early/transiently up" (only up-regulated by Mp at the 24 h time point), (2) "early/ sustained up" (up-regulated at both 24 h and 48 h after Mp contact) and (3) "late up" (only up-regulated at 48 h after Mp contact). We observed strong enrichment of hexameric motifs related to all tested promoter motifs in our sets of Mp-response genes, except for the GCC box ( Table 2). While in the promoter regions of "early/transiently" up-regulated genes only a G-box related motif seems conserved, the two remaining sets, "early/sustained up" and "late up", feature a variety of conserved motifs related to the G-box, EIN3/EIL1 binding sites, W-box, TGA box and TC-rich motifs. Thus, the up-regulation Mp-responsive genes may indeed be mediated by the combined activity of known ET/JA-associated TFs (Myc2-like, EIN3/EIL1-like) as well as SA-associated (WRKY, TGA-bZIP and TC-rich motif binding proteins) TFs. These observations are consistent with the enhanced Mp susceptibility phenotype we observed in ein3/eil1 and wrky70/54 mutants and further support that host immune responses of Arabidopsis against Mp are mediated by the combined activity of ET/JA signaling and SA signaling pathways.
Despite the fact that G box-related motifs are strongly enriched in the examined sets of Mp-induced Arabidopsis genes, we had not observed altered Mp-susceptibility of the Myc2-deficient jin1 mutant. This may reflect the complex roles of Myc2 in JA/ET-signaling, as it can suppress certain groups of genes, while activate others 56 . Lack of an Mp defense-related phenotype in jin1 may also be due to possible functional redundancy between this TF and other related Myc family members 77 . functional categories of Mp-responsive Arabidopsis genes. We further categorized all Arabidopsis genes differentially expressed after Mp based on the timing and dependency of their transcriptional changes on EIN2/JAR1 into 12 groups (labeled by roman numerals I -XII in Fig. 4E,F, Supplementary Tables S13-S25) and performed a GO term enrichment analysis. Differential gene expression changes detected by our RNA-seq analysis may reflect inducible host defense responses to combat Mp and active host responses to counter Mp-inflicted disease symptoms and damage. Furthermore, such gene expression changes may also be the consequence of Mp effector activity and serve pathogen virulence rather than host immunity and damage protection. Several general themes stand out. Consistent with our observation that known defense responses mediated by JA, ET and SA contribute to host resistance against Mp, groups of Mp-up-regulated genes are enriched for GO terms associated with responses to defense, stress, JA, ET, SA, various pathogens and, also, wounding (a known trigger of JA signaling). Similarly, up-regulated genes associated with the strongly enriched GO term "camalexin biosynthesis" are likely serving direct host immune responses, by promoting the synthesis of this indole-related phytoalexine. Host genes up-regulated to counter Mp-inflicted damage may be associated with GO terms representing protein re-folding, senescence, and responses to starvation, heat or toxic substances; all of which are particularly strongly enriched in genes up-regulated by Mp infection in the more susceptible ein2/jar1 line (Supplementary Table S25). Numerous Arabidopsis genes associated with GO terms related to plant cell wall metabolism are down-regulated by Mp infection (Supplementary Table S25). These and other gene sets related to plant growth, development, metabolism or gene expression may be actively targeted by Mp effectors to weaken the host and to counteract host defense responses.

Discussion
Using agar plate-based assays for interactions between Arabidopsis and Mp, we found the ET/JA signaling deficient ein2/jar1 double mutant to exhibit robustly enhanced susceptibility to this pathogen compared to its wild type Col-0 parental background. Three different quantitative/semi-quantitative assays (microsclerotia density, SCAR-qPCR and shoot disease index), provided highly consistent results uncovering enhanced fungal growth and host damage by Mp in this double mutant. Overall, our shoot disease index-based assay appeared to be of higher sensitivity than the root-based assays and allowed for higher throughput. Therefore, we used this assay to analyze a wider set of mutants. Using the set of assays, we also found Mp susceptibility to be enhanced in the ein2 and jar1 single mutants, which are deficient in responses to ET or JA, respectively. These results, and observations we made with additional ET-or JA-mutants, implicate ET and JA signaling as important for protection against Mp in Arabidopsis. Together, the assays can uncover multiple plant responses throughout the Mp infection process. ET-related responses are clearly evident in roots at 24 h (RNA-seq data), at 48 h (RNA-seq data and SCAR-qPCR Fig. 2B) and at 4-5 dpc (as part of the ein2/jar1 microsclerotia density increase) as well as in shoots at 7 and 14 dpc. JA-related responses also contribute to root susceptibility, for while there was more Mp tissue present in the ein2 plants than in the jar1 plants (SCAR-qPCR Fig. 2B), microsclerotia density only increased when both signaling pathways were compromised (see ein2/jar1 Fig. 2A), suggesting a possible relationship between JA signaling and microsclerotia formation that merits future investigation. In the shoot, ein2 was significantly different from ein2/jar1 at 14 dpc, though not at 7 dpc (data not shown), pointing to more of a role for JA during the later infection period.  78 . Consistent with a role for ET-and JA-signaling in Mp immunity, we also observed up-regulation of numerous Arabidopsis genes related to these processes in Mp-infected Col-0 plants. These included genes of various ERF (Ethylene Response Factor) transcription factors, JA-responsive JAZ transcriptional regulators as well as ET or JA biosynthetic enzymes. We further found promoter motifs related to known JA-and ET-response elements to be substantially enriched in Mp-responsive Arabidopsis genes. Given that Mp is considered to exhibit a mostly necrotrophic lifestyle, it is not surprising that Arabidopsis activated ET/JA-dependent immune responses upon infection with this pathogen. Both of these stress hormones are known to control plant immune responses against necrotrophic pathogens 45,46 . Our results served as critical "proof-of-concept" demonstrating that the new model pathosystem and defense assays we developed allow for the discovery of Arabidopsis processes suppressing Mp infections. Furthermore, our results provided an opportunity to use an Arabidopsis line with elevated levels of susceptibility for a comparative transcriptomics study to identify gene expression changes associated with host defense reactions against Mp.
Surprisingly, the Mp-responsive Arabidopsis transcriptome exhibited substantial overlap with gene sets known to be inducible by SA. Consistent with this, we also observed strong enrichment of promoter motifs related to binding sites of SA-associated TFs in Mp-responsive gene sets. Our Mp defense assays further confirmed a role of SA signaling in host immunity against this pathogen. Given the predominantly necrotrophic life-style of Mp, this was somewhat unexpected, as SA is known to control immune responses against biotrophic pathogens 45,79 . However, Chowdhury and co-workers 22    www.nature.com/scientificreports www.nature.com/scientificreports/ employs a short biotrophic attack strategy prior to switching to a primarily necrotrophic phase. Based on Mp marker gene expression and morphological features of Mp hyphae between 24 h and 38 h after infection of sesame, Mp seems to transition from a biotrophic phase to a necrotrophic phase. This "biotrophy-necrotrophy switch" (BNS) is accompanied by a change of physiological, biochemical and transcriptional responses of the host. Most importantly, BNS is associated with a transition from typical SA-to ET/JA-response gene expression in sesame. Our results are consistent with the observations and conclusions made by Chowdhury and coworkers.
When interpreting the timing of transcriptional changes in pathogen-infected plants, it is important to consider that infection events do not occur synchronously in all cells/tissue areas of the host. At any time point, only the "average" of multiple asynchronized events can be monitored. Thus, the apparent co-occurrence of SA and ET/JA-responsive gene expression at the two time points (24 hpc and 48 hpc) studied here may, in fact, reflect two successive defense gene expression states in the host. While each individual Arabidopsis root cell may be engaged in either pre-BNS or post-BNS response activities, the tested tissue as a whole may represent a mixture of response states. Chowdhury and co-workers made similar observations. High resolution time course studies, possibly monitoring gene expression events in individual Arabidopsis root cells, may have to be employed in the future to determine if an early pre-BNS phase in Mp is countered by SA-mediated host defenses and if later post-BNS growth of Mp induces ET/JA-dependent immunity in Arabidopsis roots.
Crop diseases caused by Mp are typically associated with drought and heat stress 16 . Thus, the induction of drought-or heat-tolerance mediating plant stress responses may be linked to immune responses against Mp. We indeed observed Mp-induced up-regulation of numerous Arabidopsis genes associated with drought and heat tolerance, such as genes encoding several DREB (Drought-Response Element Binding Protein)-type transcription factors, drought tolerance associated LEA (Late Embryogenesis Abundant) proteins, as well as various heat shock transcription factors and heat shock proteins, in the RNA-seq data. Plant tolerance to drought stress is partially controlled by the phytohormone abscisic acid (ABA). Among genes up-regulated by Mp in Col-0 were genes encoding the ABA-responsive basic helix loop helix transcription factor BHLH17, the ABA-responsive protein ABR and AFP4 (a negative regulator of ABA-responses). Future reverse genetic studies may be performed to test if ABA-signaling or other known abiotic stress response mechanisms affect the outcome of Mp-Arabidopsis interactions. In any case, our results point toward a complex interplay of multiple plant hormone-controlled signaling processes in immunity against Mp.
Having established the Arabidopsis-Mp interaction as a lab model will allow for faster progress in uncovering plant defenses against this detrimental pathogen. In this study, advantages of the Arabidopsis system became obvious. An abundance of well-characterized mutants with defined defects in signaling pathways enabled targeting of certain candidate processes (e.g. ET-and JA-signaling) by reverse genetic analyses. The existence of a well-annotated Arabidopsis genome sequence allowed for extraction of detailed information from the RNA-seq based transcriptomics experiments. Furthermore, a wealth of existing RNA-seq and microarray data was available for a wide variety of biological conditions in Arabidopsis for comparison to our results.
Next steps in the use of the new model phyto-pathosystem will include systematic testing of candidate genes identified by the RNA-seq study for their contribution to Mp defense responses using existing sequence-indexed T-DNA mutant collections [80][81][82] , which collectively cover almost 100% of Arabidopsis genes. The same reverse genetics resources can be used to test conserved QTL candidate loci potentially protecting crop species against Mp. Comprehensive collections of natural Arabidopsis accessions (ecotypes) can be tested against a panel of different Mp isolates to identify race-specific interactions with differential outcomes. Existing SNP resources for many Arabidopsis accessions or recombinant inbred lines 83 can be used to map loci responsible for the observed differential effects. Finally, Arabidopsis has proven to be an excellent system for high throughput applications, such as forward genetic screens of large populations of randomly mutagenized individuals 84 or chemical screens 69,85 . Mutations or chemicals, respectively, affecting the outcome of plant-Mp interactions, can be identified this way. We have begun some of these approaches.
Although artificial, the agar plate-based Arabidopsis root assays provided conditions for Mp to engage in its natural mode of host infection, as microsclerotia formed hyphae, which penetrated dermal root tissues resulting in colonization (as seen in soybean 38 , sesame 19 , and M. truncatula 78 ) and the formation of new microsclerotia within the host. Thus, molecular host responses identified in these assays are likely to reflect authentic processes occurring during plant-Mp interactions under natural conditions. Studying the plant-pathogen interaction on agar plates removed interference from unknown structures, compounds or organisms present in autoclaved soil that could provide additional variables affecting plant responses or pathogen behavior. The Mp/Arabidopsis experimental system established here should allow researchers to make great progress regarding the identification of key genes affecting the outcome of plant-Mp interactions. Knowledge of such genes in Arabidopsis and their orthologs in crop species will facilitate the design of new molecular markers for precise marker-based breeding approaches in economically important plants. New synthetic elicitors may also be identified that specifically activate plant immune responses active against Mp and which could serve as leads for the development of new pesticide alternatives. Beyond benefiting agriculture directly, discoveries made using our experimental pathosystem may allow for the gain of deeper insight into immune responses against Mp. The potential interplay of various hormone-pathways, which is involved in controlling immunity against Mp, is a particularly appealing subject that can be approached using our agar-plate-based Mp/Arabidopsis infection assays.
Mp (originally isolated from UCR Ag Operations Field 11 by members of Philip Roberts' laboratory 16 ) was propagated by adding a plug to potato dextrose agar (BD Difco TM , http://us.vwr.com) -containing plates (10 ml per plate), incubating at 34 °C for 10 days and then allowed to dry at RT for at least four weeks.

Mp infection assays.
Mp infection plates were created by adding Mp inoculum (approximately 2000 microsclerotia in 2-3 ml media for microsclerotia counting and RNA-seq roots and twice the amount for the shoot disease index assay) to plates, placing plates in a dark incubator at 34 °C for three days and then transferring the plates to growth room conditions for an additional three days. The media used to pour plates and create inoculum contained ½ MS (Murashige and Skoog), 0.05% MES, 0.25% sucrose and 0.87% agar; pH was 5.7 prior to the addition of agar. Dried Mp propagation plate contents were ground in 10-20 ml sterile water with mortar and pestle, added to media and counted in 100μl drops on slides until the desired concentration was reached.
For the microsclerotia density assay, 10 day-old seedlings were gently transferred to Mp infection plates, with 12-14 plants per plate, and these plates were placed vertically in growth room conditions for four to five days. To arrest Mp development, 20-30 ml 75% EtOH was added to each plate. Microsclerotia were counted under bright field microscopy and root lengths were measured using ImageJ 100 . Magnified images were taken using a Leica DM/LB2 (Leica, Wetzlar, Germany) microscope equipped with an RT colour SPOT camera. At least three biological replicates were conducted with a minimum of three plates per line. Statistical significance relative to Col-0 was determined by Student's t-test, p < 0.05.
For the Mp-Arabidopsis shoot disease index, 10 day-old seedlings were gently transferred to Mp infection plates, with two lines per plate, 10 plants per line (n = 30), and these plates were placed horizontally in growth room conditions for two weeks. Images were taken of whole plates throughout the infection period to monitor progress of disease symptoms in plants. Individual plants were assessed according to a disease index scoring system of "0", for healthy plants, through "5", for dead plants (Fig. 3B). At least three biological replicates were conducted and significance relative to Col-0 was determined using the Wilcoxon Rank Sum test for two independent samples by Mathcracker on https://mathcracker.com/wilcoxon-rank-sum#results.
Transcriptome profiling by mRNA-seq. For RNA-seq, 10 day-old seedlings were transferred to agar plates without Mp or to Mp infection plates, with 12-14 plants per plate, and these plates were placed vertically in growth room conditions for 24 h or 48 h. Plant roots were separated from shoot tissue using a blade and flash frozen in liquid nitrogen at 24 h or 48 h post transfer. Total RNA was isolated from roots using TRIzol (Invitrogen TM , http://www.thermofisher.com). RNA was processed (74204 QIAGEN, http://www.qiagen.com, and AM1907 Invitrogen TM , http://www.thermofisher.com) and libraries were prepared with the NEBNext Ultra Directional RNA Library Prep Kit for Illumina by following the manufacturer's instruction (E7490S, E7335S, E7420S, New England Biolabs, http://www.neb.com). Root tissues were separately analyzed for each line (Col-0 or ein2-1/jar1-1), treatment (+/− Mp) and time point (24 h or 48 h). Two independent biological replicates, with three technical replicates for each experimental condition, were performed.
Libraries were pooled and sequenced on an Illumina NextSeq. 500 (Illumina, San Diego, CA, USA) platform at the UCR Genomics Core Facility. Reads that passed Illumina's quality control filters were further processed. The quality of sequencing reads was assessed using FastQC v 0.11.5 (http://www.bioinformatics.babraham.ac.uk/ projects/fastqc/). Unique reads were mapped to the Arabidopsis genome (TAIR10) using STAR v 2.5.3a 101 with default settings and a known splice site file, built from Araport annotation file v11. Reads in gene regions were counted using featureCounts 102 . The expression fold-change of each gene was calculated using the R package DESeq. 2 v1.14.1 103 with the threshold for differentially expressed genes set to p value < 0.05.
The MA-plots were generated with the function plotMA in DESeq. 2 and illustrate log-fold change (M-values, i.e. the log of the ratio of level counts for each gene between two samples) against the log-average (A-values, i.e. the average level counts for each gene across the two samples). They showed the log 2 fold changes attributable to a given variable over the mean of normalized counts for the compared samples. Differentially expressed gene points were red if the adjusted p value was less than 0.01. Genes with similar expression levels in two samples appeared around the horizontal line y = 0. Data comparison AGIs can be found in Supplementary Tables S5-S12. Comparisons between experimental gene sets and sets of genes responding to other stimuli were done using Venny 2.1, (http://bioinfogp.cnb.csic.es/tools/venny/index.html).