Pattern‐triggered immunity restricts host colonization by endophytic fusaria, but does not affect endophyte‐mediated resistance

Abstract Fusarium oxysporum (Fo) is best known as a host‐specific vascular pathogen causing major crop losses. Most Fo strains, however, are root endophytes potentially conferring endophyte‐mediated resistance (EMR). EMR is a mechanistically poorly understood root‐specific induced resistance response induced by endophytic or nonhost pathogenic Fo strains. Like other types of induced immunity, such as systemic acquired resistance or induced systemic resistance, EMR has been proposed to rely on the activation of the pattern‐triggered immunity (PTI) system of the plant. PTI is activated upon recognition of conserved microbe‐associated molecular patterns (MAMPs) of invading microbes. Here, we investigated the role of PTI in controlling host colonization by Fo endophytes and their ability to induce EMR to the tomato pathogen Fo f. sp. lycopersici (Fol). Transgenic tomato and Arabidopsis plants expressing the Fo effector gene Avr2 are hypersusceptible to bacterial and fungal infection. Here we show that these plants are PTI‐compromised and are nonresponsive to bacterial‐ (flg22) and fungal‐ (chitosan) MAMPs. We challenged the PTI‐compromised tomato mutants with the EMR‐conferring Fo endophyte Fo47, the nonhost pathogen Fom (a melon pathogen), and with Fol. Compared to wild‐type plants, Avr2‐tomato plants became hypercolonized by Fo47 and Fom. Surprisingly, however, EMR towards Fol, induced by either Fo47 or Fom, was unaffected in these plants. These data show that EMR‐based disease resistance is independent from the conventional defence pathways triggered by PTI, but that PTI is involved in restricting host colonization by nonpathogenic Fo isolates.

Besides genetic resistance in the host, biocontrol is an alternative method to control wilt diseases. Many studies have shown that endophytic Fo strains can confer protection to pathogenic Fo isolates (de Lamo & Takken, 2020). A potential advantage of this method is that protection does not rely on single R genes and hence might be more durable and more difficult to overcome by the pathogen. A caveat of biocontrol is that the level of resistance is typically not as high as that conferred by resistance genes (de Lamo et al., 2018). Fo-based biocontrol is the sum of different components. The first component consists of the endophyte directly mycoparasitizing or competing with the pathogen for nutrients and root entry points, limiting progression of the latter. Another component encompasses a root-specific immune response triggered by Fo endophytes (de Lamo & Takken, 2020), and we refer to this response as endophyte-mediated resistance (EMR).
In tomato, EMR is induced independent of the defence hormones jasmonic acid (JA), ethylene (ET), and salicylic acid (SA) (Constantin et al., 2019). The colonization pattern of wild-type and hormone mutants by the Fo47 endophyte is identical and also EMR is unaffected in tomato plants impaired in SA accumulation (NahG), JA biosynthesis (def1), or ET-production (ACD) and -sensing (Nr) (Constantin et al., 2019). The independence of EMR on SA, JA, and ET implies that EMR might be distinct from the known induced immune responses that do rely on either SA (systemic acquired resistance, SAR) or JA/ET (induced systemic resistance, ISR) (Fu & Dong, 2013;Pieterse et al., 2014). These latter responses are preceded by a PTI response, and therefore we set out to investigate whether PTI is involved in EMR and/or colonization of the host by Fo.
To study the link between EMR and PTI we used tomato as model.
As a PTI tomato mutant we employed a transgenic line that expresses the Fol effector gene Avr2. Avr2-expressing plants show compromised PTI responses such as growth inhibition, ROS production, and callose deposition upon treatment with the bacterial MAMP-derivative flg22 (Di et al., 2017). Lines expressing Avr2 are also hypersusceptible to bacterial and fungal pathogens, suggesting that besides bacterial MAMP-induced PTI the plant responses to fungal MAMPs are also compromised. Here, we describe that in Avr2 plants PTI triggered by the fungal MAMP chitosan is compromised as well.
As an established model system for EMR, the tripartite interaction between tomato, Fo47, and Fol, was used (Aïcha et al., 2014;Aimé et al., 2013;Bolwerk et al., 2005;Constantin et al., 2019;de Lamo et al., 2018). Besides in tomato, Fo47 confers resistance to Fo pathogens in various plant species, such as asparagus (Asparagus officinalis; Elmer, 2004), flax (Linum usitatissimum; Trouvelot et al., 2002), and eucalyptus (Eucalyptus viminalis; Salerno et al., 2000), and to the vascular pathogen Verticillium dahliae and root-infecting oomycetes Phytophthora capsici in pepper (Capsicum annuum; Veloso & Díaz, 2012) and Pythium ultimum in cucumber (Benhamou et al., 2002). Unlike Fol, Fo47 is a poor root-and stem colonizer and stem colonization is typically restricted to the crown and although it effectively colonizes roots the amount of fungal biomass is approximately 20-fold lower than that of a pathogenic strain . Notably, whereas stem and root colonization by Fol are reduced by about 9-fold upon coinoculation with Fo47, the endophyte becomes a better stem colonizer in the presence of the pathogen despite not colonizing the roots more extensively . Besides Fo47, which has a purely endophytic lifestyle, we also included in our bioassays the nonhost pathogen Fo f. sp. melonis (Fom), which typically infects melon. Together these experiments allowed us to investigate the role of PTI in EMR, and in colonization of tomato by the endophyte Fo47 and by Fom.

| Endophyte-mediated resistance can be conferred by endophytic and nonhost isolates of F. oxysporum
Besides Fo endophytes such as Fo47, nonhost Fo pathogenic strains are also reported to confer EMR (Biles & Martyn, 1989;Díaz et al., 2005). Nonetheless, it is unknown whether nonhost Fo pathogens confer EMR in tomato to an extent similar to endophytic strains. To address this question, the EMR-inducing potential of a melon-pathogenic strain, Fom001, was assessed in tomato and compared to that of Fo47. The latter is a well-known EMR inducer when coinoculated with Fol (Constantin et al., 2019;de Lamo et al., 2018). Ten-day-old MoneyMaker (MM) seedlings were root dip-inoculated with water (mock), Fo47, the tomato pathogen Fol4287 or Fo47:Fol, a mixture of Fo47 and Fol4287 ( Figure 1a). Three weeks postinoculation (wpi) Fo47 was found to confer EMR as fresh weight (FW) of plants coinoculated with Fo47 and Fol4287 was significantly higher than that of plants infected with Fol4287 alone (Figure 1b). In addition, the disease index (DI) was reduced when the pathogenic and Fo47 biocontrol F I G U R E 1 Both Fo47 and the melon pathogenFusarium oxysporum(Fo) f. sp.melonis(Fom001) confer endophyte-mediated resistance (EMR) to Fo f.sp.lycopersici(Fol4287) in tomato. (a) Ten-day-old MoneyMaker seedlings were root dip-inoculated with water (mock), Fo47, Fom001, Fol, a mixture of Fo47 and Fol4287 (Fo47:Fol), or Fom001:Fol. In this figure a minimum of 10 seedlings was used for Fo47 treatment while 25 seedlings were used for the Fom treatments. (b) Three weeks postinoculation fresh weight (FW) and (c) disease index (DI) were scored. The experiment was repeated three times with 10, 15, or 25 plants/treatment showing similar results. (d), (e), and (f) The same bioassays were carried out with Fom001. An unpaired comparison for FW and DI was performed using the nonparametric Mann-WhitneyUtest (*p < .05, **p < .01, ****p < .0001) strains were coinoculated as compared to infection solely with the pathogenic strain ( Figure 1c). Using the same procedure Fom001 was found to confer EMR to Fol4287 to an extent similar to Fo47 ( Figure 1d) as FW and DI were affected to a comparable extent in this tripartite interaction (Figure 1e,f). These findings show that the nonhost pathogen Fom001 is an endophyte of tomato and confers EMR, reducing disease symptoms caused by a tomato pathogenic Fo strain.

| Avr2 compromises tomato responsiveness to a fungal MAMP
A tomato PTI-mutant is instrumental to assess involvement of the PTI response in EMR. However, because PTI knockout mutants have not been described in tomato, we investigated whether the transgenic line MM-∆spAvr2-30 could be used as a proxy. This transgenic line accumulates Avr2 intracellularly, resulting in a compromised PTI response as exemplified by its reduced responsiveness to the bacterial MAMP flg22 (Di et al., 2016(Di et al., , 2017. If Avr2 targets a general component of PTI signalling, and not a flg22-specific factor, then this line provides an excellent proxy of a PTI mutant to study the role of PTI in EMR in tomato. To test whether Avr2 affects the responsiveness of tomato to fungal MAMPs, leaves of MM and MM-∆spAvr2-30 were infiltrated with the fungal MAMP-derivative chitosan and subsequently callose depositions were monitored. Chitosan was used instead of chitin (poly-N-acetylglucosamine) as this deacetylated variant of chitin is more soluble and has been reported to induce a potent defence response in tomato (Rendina et al., 2019). As a positive control for PTI induction flg22 was used while Milli-Q water served as negative control. Formation of callose deposits was detected in chitosan-and flg22-treated leaves to a significantly higher extent than in water-infiltrated leaves ( Figure 2). This result shows that these depositions F I G U R E 2 Compared to wild-type MoneyMaker (MM) tomato, MM-∆spAvr2-30plants show a reduced number of callose depositions in leaves upon flg22 and chitosan infiltration. (a) flg22 (100 nM)-and chitosan (100 µg/ml)-induced callose depositions were visualized by aniline blue staining in MM and MM-∆spAvr2-30leaves. (b) Quantification of flg22-and chitosan-induced callose deposits per mm 2 . Three MM plants and three MM-∆spAvr2-30plants were selected for flg22 or chitosan infiltration. Four leaf discs were taken per plant. For the mock treatments two plants were infiltrated. Three pictures were taken from different areas of the treated leaf discs. The experiment was replicated twice with the same results. An unpaired comparison was performed using the nonparametric Mann-WhitneyUtest (****p < .0001) are a specific response to the MAMPs and are not due to the infiltration procedure. As expected, MM-∆spAvr2-30 leaves infiltrated with flg22 showed significantly fewer callose depositions as compared to wild-type MM leaves, confirming previous reports (Figure 2b;Di et al., 2017). Compared to wild-type MM, MM-∆spAvr2-30 leaves infiltrated with chitosan showed significantly fewer callose depositions, indicating that Avr2 also suppresses chitosan-induced defence responses.
As mentioned, chitosan is a deacetylated form of chitin and this modification could affect receptor-specific perception by the plant.
To assess whether the chitosan preparation used in our studies triggers a receptor-specific PTI response, Arabidopsis Col-0 wild-type and the cerk1-2 mutant were treated with chitosan and the formation of callose deposits was monitored. We observed that callose depositions were only induced in wild-type Col-1 plants, but not in the cerk1-2 mutant, indicating that the preparation contains a specific MAMP recognized by the CERK1 receptor ( Figure S1a). As a positive control for PTI-inducibility flg22 was used, and found to trigger a callose deposition response in both wild-type and cerk1-2 mutant plants ( Figure S1b). To test whether Avr2 targets a conserved component in chitosan-and flg22-induced PTI signalling the callose deposition response in ∆spAvr2 transgenic Arabidopsis was analysed ( Figure S1c).
Compared to the wild-type Col-0 plants a significantly reduced callose deposition was observed in the ∆spAvr2 transgenic Arabidopsis upon chitosan and flg22 treatment ( Figure S1). The observations that flg22-and chitosan-induced PTI responses are compromised in both MM-∆spAvr2-30 tomato and in ∆spAvr2 Arabidopsis indicate that the Avr2 effector targets a conserved-and shared-signalling component downstream of the respective PRR receptors, making ∆spAvr2 transgenic lines a good proxy for a PTI mutant.

| Endophytic F. oxysporum strains hypercolonize PTI-compromised tomato
To establish themselves within the roots, endophytes are hypothesized to suppress and/or avoid activation of extensive host defences (Teixeira et al., 2019). To investigate whether PTI affects colonization by endophytic Fo of tomato, stem pieces from MM, MM-Avr2-7, and MM-∆spAvr2-30 tomato plants were harvested 3 wpi with Fo47 (10, 15, or 25 plants/tomato genotype). Stem sections where taken at crown, cotyledon, and the first true leaf to monitor the extent of host colonization by the fungus (Figure 3a). Line MM-Avr2-7 was included as a negative control in which PTI is not suppressed. In this line Avr2 is secreted into the apoplast while its virulence function has been reported to be intracellular (Di et al., 2017). Altogether, an effective PTI response to Fo47 appears to be instrumental in controlling the extent of host colonization by a Fo endophyte as MM-∆spAvr2-30 plants are hypercolonized, as compared to wild-type and MM-Avr2-7 plants.
Next, we set out to quantitatively assess whether Fo47 colonization of stems correlated with a more extensive colonization of roots.
Thereto genomic DNA was isolated from roots of inoculated wildtype, MM-Avr2-7, and MM-∆spAvr2-30 plants. Subsequently, quantitative PCR (qPCR) was performed with a Fo47-specific primer pair to quantify fungal biomass relative to tomato β-tubulin, which served as endogenous control (Figure 3d). Fo47-inoculated MM wild-type and MM-Avr2-7 plants consistently showed the least colonization, while MM-∆spAvr2-30 showed a significantly higher amount of fungal biomass in the three tomato lines analysed. These data show that a compromised PTI results in a more abundant colonization of tomato roots by the endophyte.
To assess whether PTI also restricts colonization of a nonhost pathogen, the MM, MM-Avr2-7, and MM-∆spAvr2-30 tomato lines were inoculated with Fom001. Like Fo47, Fom001 was found to colonize MM-∆spAvr2-30 to a higher extent than wild-type MM and MM-Avr2-7 plants. Colonization of the latter two genotypes was highly similar (Figure 3e,f), resembling the pattern observed for Fo47.
Together, these experiments show that both Fo47 and Fom001 become hypercolonizers of PTI-compromised tomato plants.

| Endophyte-mediated resistance is unaffected in PTI-compromised tomato
Previous studies showed that endophytes can reduce host susceptibility to different pathogens (Ghorbanpour et al., 2018;de Lamo & Takken, 2020). We were interested in testing whether PTI affects EMR conferred by either Fo47 (Figure 4) or the melon pathogen Fom001 ( Figure 5). MM, MM-Avr2-7, and MM-∆spAvr2-30 tomato plants were inoculated with water (mock), Fo47, Fol4287, or a 1:1 mixture of both (Figure 4a). Fo47 was found to confer EMR to Fol4287 in the three different tomato genotypes as FW of coinoculated plants was always significantly higher than that of solely

| D ISCUSS I ON
We report that the endophyte Fo47 and the melon pathogen Fom001 become hypercolonizers of PTI-compromised tomato plants. Surprisingly, the biocontrol conferred by both Fo strains towards Fol4287 was unaffected in these mutant plants. Our findings imply that EMR operates independently from PTI signalling. As proxy for a PTI mutant we used the transgenic MM-∆spAvr2-30 line that produces the fungal effector protein Avr2, compromising PTI signalling. Previously, these plants have been reported to be hypersusceptible to various pathogens, including bacterial and fungal pathogens, and to exert a compromised immune response on treatment with a bacterial MAMP (Di et al., 2017). In this study we extend these findings by reporting that ∆spAvr2 tomato, but also ∆spAvr2 F I G U R E 3 Intracellular presence of Avr2 in tomato facilitatesFusarium oxysporum(Fo) endophytism. (a) To monitor stem colonization by Fo endophytes stem sections from Fo47-or Fom001-inoculated MoneyMaker, MM-Avr2-7, and MM-∆spAvr2-30were harvested at the crown, cotyledon, and first true leaf levels. Stems sections of Fo47 (hygromycin-resistant)-infected plants were surface-sterilized and placed on (b) potato dextrose agar supplemented with hygromycin. Plates were scanned after 4 days of incubation at 25 °C in the darkness. In the pictures, a subset of five replicates out of a total of around 20 is shown. (c) Positive fungal outgrowth was quantified. Theyaxis represents the percentage of biological replicates that were (non-)colonized to the crown, cotyledon, or first true leaf level. One representative bioassay is shown as illustration. (d) Root colonization by Fo47 was quantified by means of quantitative PCR. Three technical replicates were done for each biological replicate. (e) and (f) Stem colonization of Fom001 in the same tomato lines was also monitored. The experiments were replicated twice. An unpaired comparison for the relative quantification and the fungal stem reisolations was performed using the nonparametric Mann-Whitney U test (*p < .05, **p < .01, ***p < .001, ****p < .0001) Arabidopsis plants, are also compromised in their response to the fungal MAMP chitosan.

Hypercolonization of stems and roots of ∆spAvr2 plants by
Fom001 and Fo47 shows that PTI controls the extent of host colonization by nonpathogens. Hence, the hypothesized suppression and/or avoidance of activation of host defences by endophytes to colonize roots (Teixeira et al., 2019) does not seem to involve a complete suppression of PTI. Indeed, Fo47 infection has been reported to transiently induce immune signalling in tomato roots, resulting in papilla formation compromising host cell penetration by Fo47 (Benhamou et al., 2002). Our observation that PTI controls root colonization of Fo47 is interesting in the light of the recent finding that tissue damage is required for differentiated Arabidopsis roots to mount a full PTI response (Zhou et al., 2020). Spontaneous cell death, or cell death induced by the formation of lateral roots, nematode feeding, or pathogen presence, triggers PRR expression in F I G U R E 4 Fo47 does not lose its endophyte-mediated resistance-inducing capabilities in pattern-triggered immunity-compromised plants. (a) Ten-day-old MoneyMaker, MM-Avr2-7, and MM-∆spAvr2-30seedlings were root dip-inoculated with water (mock), Fo47, Fol, or a mixture of Fo47:Fol4287. A minimum of 10 seedlings was used for the mock treatment and 20 were used for the other treatments. Three weeks postinoculation (b) fresh weight (FW) and (c) disease index (DI) were scored. The experiment was repeated three times with similar results. An unpaired comparison for FW and DI was performed using the nonparametric Mann-Whitney U test (*p < .05, **p< .01, ***p < .001, ****p < .0001) neighbouring cells, enabling these cells to perceive MAMPs and mount a PTI response. Whether Fo-induced host cell death is required for tomato roots to mount PTI remains a question for future study. The observation that Fo47 induces cell death near infection sites (Alabouvette et al., 2009), while pathogenic strains do not (Alabouvette et al., 2009;He et al., 2002;Humbert et al., 2015;Olivain et al., 2003), would be consistent with our finding that PTI is involved in restricting root colonization by the endophyte. Of note, although both Fom001 and Fo47 become hypercolonizers of PTI suppressed plants, these strains did not become pathogenic and did not cause disease symptoms. A possible explanation for this is that they lack the specific effector repertoire of Fol that is located on its pathogenicity chromosome. Transfer of this specific chromosome to Fo47 converts it into a genuine tomato pathogen, showing that this chromosome is required and sufficient for pathogenicity (Ma et al., 2010). Apparently, these Fol effectors have functions besides suppressing PTI that are required for pathogenicity and to cause disease on tomato.

F I G U R E 5
Melon pathogen Fom001 does not lose its endophyte-mediated resistance-inducing capabilities in pattern-triggered immunity-compromised plants. (a) Ten-day-old MoneyMaker, MM-Avr2-7, and MM-∆spAvr2-30seedlings were root dip-inoculated with water (mock), Fom001, Fol,or a mixture of Fom001:Fol4287. Ten MM-∆spAvr2-30and 15 MM or MM-Avr2-7plants were used per treatment. Three weeks postinoculation (b) fresh weight (FW) and (c) disease index (DI) were scored. The experiment was repeated three times. An unpaired comparison for FW and DI was performed using the nonparametric Mann-WhitneyUtest (**p < .01, ****p < .0001) The observation that EMR is still induced in ∆spAvr2 plants can be explained by different nonexclusive models. One model identifies a plant-directed response as the main factor in disease reduction, while the other proposes antagonism between the fungal strains as the main contributor. The latter scenario fits the interaction be-  (Olivain et al., 2006). In addition, as relatively low doses of Fo47 inoculum confer protection to Fol-induced disease a plant-mediated response is likely to be involved (Fravel et al., 2003). In our study a 1:1 ratio of endophyte and pathogen was used, but equally strong protection was found using a 1:10 ratio and even at a 1:100 ratio disease development was suppressed by the endophyte . So, although direct competition between the strains might contribute to EMR, the host will certainly contribute to the observed resistance in this tripartite system.
Whereas our data strongly suggest that PTI is not involved in EMR, its involvement cannot completely be excluded as PTI is compromised but not fully absent in the MM-∆spAvr2-30 and the ∆spAvr2 Arabidopsis lines. Chitosan-induced callose depositions are, for instance, not completely absent in ∆spAvr2 plants (Figures 2 and   S1) as is phosphorylation of MAPKs upon flg22 treatment (Di et al., 2017). Completely excluding potential involvement of PTI in EMR awaits generation of stable tomato knockout mutants in which PTI signalling is practically absent, like the bbc Arabidopsis mutants (Xin et al., 2016). However, other lines of evidence also imply that PTI is not instrumental for EMR induction. PTI induction typically results in induction of expression of defence genes such as PR-1 and the encoded protein is not found in xylem sap of tomato plants coinoculated with Fo47 and Fol (de Lamo et al., 2018). The absence of this PTI marker suggests that either PTI is not, or is only transiently, in-  (Constantin et al., 2019). Together with the observation that EMR is still induced in tomato lines that are compromised in either JA, ET, or SA signalling (Constantin et al., 2019), these data provide an argument for EMR being distinct from the known induced resistance responses that rely on these hormones and are triggered by PTI (Mishina & Zeier, 2007).
Taken together, our data support a hypothesis in which Foinduced EMR represents a type of induced resistance that is distinct from the other induced resistance responses in plants, such as SAR and ISR, as it is still functional in tomato plants in which PTI signalling is compromised. This response seems specifically potent to control wilt disease caused by pathogenic Fo strains and to root-invading pathogens (de Lamo & Takken, 2020). A detailed molecular understanding of the mechanisms underlying in Fo-based EMR will be instrumental in developing novel biocontrol strategies to limit our dependency on pesticides in agriculture.

| Plant and fungal materials and cultivation conditions
Three different tomato lines were used that are susceptible to its signal peptide resulting in a cytoplasmically localized Avr2 protein (Di et al., 2016). Besides wild-type Col-0, an ∆spAvr2-23 line (Di et al., 2016) and a cerk1-2 mutant (Miya et al., 2007) were used. Fungal infections were carried out with Fo strains carrying a gene that confers resistance to hygromycin. The endophytic strains used were Fo47 (FP1544) and Fom001 (FP1577); the latter is a pathogen on melon. As pathogenic strain, Fol4287 was used. Tomato plants were grown in a climate-controlled greenhouse at 25 °C, 65% relative humidity and a 16 hr photoperiod. Arabidopsis plantlets were grown under short day conditions, 13 hr/11 hr dark/light cycles at 22 °C.

| Chitosan and flg22 preparation
Chitosan with a low molecular weight (50-190 kDa, 75%-85% deacetylated; Sigma-Aldrich) was prepared as described (Rendina et al., 2019). In brief, chitosan was dissolved in 0.2 M acetic acid to a concentration of 1 g/ml, stirred overnight, and diluted in Milli-Q water to a concentration of 5 mg/ml. Flg22 was dissolved in sterile Milli-Q water at a concentration of 100 μM.

| Callose deposition assays and microscopy
Leaves of 3-week-old tomato plants or 4-week-old Arabidopsis plants were syringe-infiltrated with water, flg22 (100 nM), or chitosan (100 μg/ ml) and after 24 hr leaf discs were taken from the infiltrated areas.
Staining was done for 60-120 min with a 0.01% aniline blue solution (dissolved in 0.07 M sodium phosphate buffer) at pH 9. The samples were mounted in 50% glycerol and callose was visualized using a Leica MZ FLIII fluorescence microscope with a DAPI filter (UV filter, excitation 360 nm, emission 420 nm). The number of callose depositions within a frame (5× or 8× magnification) was counted using Fiji (https:// imagej.net/Fiji). A Mann-Whitney U statistical test was applied on the number of callose deposits in tomato and a one-way analysis of variance (ANOVA) was used for the number of Arabidopsis deposits using the software PRISM 7.0 (GraphPad).

| Fusarium inoculation assays
Hygromycin-resistant Fo47, Fom001, or Fol4287 were inoculated from glycerol stock to potato dextrose agar (PDA) plates supplemented with 100 mg/L hygromycin. Plate cultures were grown for at least 5 days at 25 °C in darkness. Agar plugs were used to inoculate 100 ml minimal medium (0.17% yeast nitrogen base without amino acids or ammonium sulphate, 3% sucrose and 100 mM KNO 3 ).
Cultures were incubated in the dark at 25 °C and 150 rpm for 5 days.
Thereafter, cultures were filtered through Miracloth (Calbiochem) and diluted to generate a microconidial inoculum of 10 7 spores/ml (Di et al., 2016;de Lamo et al., 2018). Coinoculum of endophyte (Fo47 or Fom001) and pathogen (Fol4287) was prepared in a 1:1 ratio (10 7 microconidia/ml each). Subsequently, 10-day-old tomato seedlings (MM, MM-Avr2-7, and MM-∆spAvr2-30) were uprooted and the roots were trimmed at 1 cm to promote fungal infection. Seedlings were root dip-inoculated by placing them for 5 min in water (mock) or a microconidial suspension of the endophyte (Fo47 or Fom001), pathogen (Fol4287), or a mixture of both. Seedlings were repotted and 3 weeks postinoculation (wpi) fresh weight (FW) and disease index (DI) were scored as described (de Lamo et al., 2018). A Mann-Whitney U test was applied on the FW and DI using the software PRISM 7.0 (GraphPad).

| Fungal stem reisolation assays
Three weeks after inoculation stem pieces were collected and analysed for fungal presence. Stem pieces were surface sterilized with 70% ethanol for approximately 3 min and subsequently rinsed with sterile water. From each stem piece, one slice (c.0.5 cm thick) was cut from the crown, cotyledon, and first true leaf segment ( Figure 3a) and placed on PDA plates supplemented with penicillin (100 μg/ml) and streptomycin (200 μg/ml) to prevent bacterial growth. Additionally, the PDA plates contained hygromycin (100 μg/ml) to select for Fo.
The plates were kept at 25 °C for 4 days and then scanned using an EPSON Perfection V800 Photo scanner and processed with the software package SilverFast 8 (LaserSoft Imaging). Thereafter fungal outgrowth was analysed. For statistical analyses, a 0 was given to noncolonized plants, 1 for plants that were colonized until the crown, 2 for colonization until the cotyledon and 3 for colonization until the first true leaf level. These categories were analysed for statistical differences using a Mann-Whitney U test.

| Relative quantification of Fusarium root colonization by qPCR
Three wpi roots from Fo47-inoculated MM, MM-Avr2-7, and MM-∆spAvr2-30 plants were harvested and snap-frozen into liquid nitrogen. Root tissue was thoroughly ground in a mortar and genomic DNA (gDNA) isolation was performed with the GeneJET Plant Genomic DNA Purification Mini Kit (Thermo Scientific). The isolated gDNA was used for real-time quantitative PCR (qPCR). Each qPCR contained 1 μl of gDNA 865.14.003.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.