The B. distachyon WT ecotype Bd21-3 (Vogel et al., 2009) and all B. distachyon lines generated or selected in this study (Table 1) were cultivated as described in Pasquet et al. (2016). For expression studies, roots and leaves were collected from three week-old plants whereas spikes were collected from 5 week-old plants (mid-anthesis stage plus 96 h that is the same stage as the one chosen for expression analysis of defense genes). For hydroponic cultures, the palea and lemma of each seed were removed and seeds were surface sterilized by incubation in a 0.6% sodium hypochlorite solution for 5 min with gentle shaking followed by three rinses in sterile distilled water. Sterilized seeds were subsequently incubated for 5 days at 4°C in sterile distilled water then pre-germinated on filter paper soaked with sterile water for three days at 24°C, both steps in the dark. Seedlings were transferred on hydroponic boxes (Araponics system, http://www.araponics.com/) in sterile self-made liquid 14 Murashige and Skoog medium [2.1 mM NH₄NO₃, 1.9 mM KNO₃, 0.3 mM CaCl₂.2H₂O, 0.15 mM MgSO₄.7H₂O, 10 μM H₃BO₄, 10 μM MnSO₄.H₂O, 0.5 μM KI, 3 μM ZnSO₄.7H₂O, 0.1 μM Na₂MoO₄.2H₂O, 0.01 μM CoCl₂.6H₂O, 0.01 μM CuSO₄.H₂O, 10 μM FeCl_2, 1 mM KH₂PO₄, (Murashige and Skoog, 1962)] for 4 weeks, with the medium changed every 3 days. One week before recovery of exudates for SL detection and quantification assays, plants were transferred to sterile self-made phosphate deficient (-KH₂PO₄) 14 MS.

The BdCYP711A29 complementary DNA (cDNA) was amplified from spikelet cDNAs using primers BdCYP711A29_BI5’ATG and BdCYP711A29_EV3’TAA (Supplementary Table 1), adding a BamHI restriction site at the 5’-end and an EcoRV restriction site at the 3’-end, respectively. The PCR product was digested using the BamHI and an EcoRV restriction enzymes, purified using a NucleoSpin^® Gel and PCR Clean-up kit (Macherey-Nagel EURL, Hoerdt, France) using the manufacturer’s instructions then ligated into the pENTR1A plasmid linearized by the same restriction enzymes. The resulting pEntry-CYP711A29 plasmid was used to transfer the BdCYP711A29 cDNA fragment into the pIPKb002 binary vector (Himmelbach et al., 2007) by in vitro recombination using the Gateway^® LR Clonase^® II Enzyme mix according the manufacturer’s recommendations (Invitrogen^TM, Life Technologies SAS, Saint-Aubin, France). The resulting construct, named pIPKb002::BdCYP711A29, carried both the BdCYP711A29 full-length cDNA under the control of the Zea mays ubiquitin promoter (ZmUbi) and a hygromycin resistance expression cassette allowing selection of the primary transformants.

The pIPKb002::BdCYP711A29 binary vector was then electroporated into Agrobacterium tumefaciens (AGL1 strain). The Bd21-3 WT line was genetically transformed using a method adapted from that described by Vogel and Hill (2008) and Alves et al. (2009). Selection of the transformants and segregation analysis were conducted as described in Pasquet et al. (2016).

Two thousand five hundred mutant families the B. distachyon TILLING mutant collection¹ available at the Institute of Plant Sciences Paris Saclay (Orsay, France) were screened for point mutations via Illumina^® sequencing of a 401-bp fragment located at the 3′ end of the BdCYP711A29 gene encompassing a region encoding important domain of the C-terminal part of the protein: the P(E)R(F) signature, the EER triad and the heme binding domain (Supplementary Figures 1, 2). Primers used for the generation of the BdCYP711A29 PCR product were CYP711A29-F1 and CYP711A29-R (Supplementary Table 1).

Fusarium graminearum strain PH-1 (FgDON⁺) and the Δtri5 mutant strain MU102 (FgDON^–; Cuzick et al., 2008) unable to produce DON were cultured as described in Pasquet et al. (2016).

An average of 1,000 F. graminearum macroconidia were plated on water agar slides containing either 0.01% DMSO or various concentrations of (rac)-orobanchol (OlChemIm, Olomouk, Czech Republic) diluted into 0.01% DMSO. Slides were further incubated at 26°C in a moist environment for 12 h. Germ tubes were then counted on a minimum of 500 macroconidia per condition per replicate through observation on an Axio Zoom V.16 under bright field light (Zeiss, Marly-le-Roi, France). To determine the impact of exudates of different B. distachyon lines on F. graminearum macroconidial germination, macroconidia at a final concentration of 10⁵ macroconidia/mL were directly incubated in tubes containing 2 mL of each recovered exudate. After incubation for 12 h at room temperature, the germination percentage was determined on a minimum number of 1,000 macroconidia. For each experiment, three biological replicates were performed.

For DON application, a single floret per spike at mid-anthesis was inoculated by a mixture containing 2 μg of the myxotoxin (Sigma-Aldrich, Saint-Quentin Fallavier, France) in 5 μL of a mixture of acetonitrile (1.65 μL) and 0.01% Tween 20 (1.35 μL). The control condition corresponded to spikes inoculated with the same mixture without the mycotoxin. For fungal infection, whole spikes at mid-anthesis were sprayed with the fungal spore suspension (1 × 10⁵ conidia/mL), until dripping. Inoculated plants were covered with clear plastic bags for which the internal face had been sprayed with distilled water beforehand. The first 24 h inoculated heads were kept in the dark, then incubated with a photoperiod of 16 h light and 8 h darkness at 20°C with the same light intensities as those used for plant development (Pasquet et al., 2016). Applications of 0.01% Tween 20 was performed as control condition for each inoculation experiment. Symptoms were observed at 7 and 14 days after spraying of the conidial suspension. A spikelet was considered as symptomatic if at least half of its florets were symptomatic.

Leaves from three 2-week old plants plant or five spikelets from independent plants were ground in liquid nitrogen and total RNA was extracted from 0.1 g of the resulting powder using TRIzol^® (Invitrogen, Life Technologies SAS, Saint-Aubin, France) followed by an RNase-free DNase I step (Ambion^®, Applied Biosystems, Courtaboeuf, France) according to manufacturers’ instructions. Total RNA was further purified using the NucleoSpin RNA Clean-up XS kit (Macherey-Nagel, Hoerdt, France).

For quantification of F. graminearum DNA, 10 spikes spray-inoculated with either of the two strains used in this study were pooled per time-point. Genomic DNA was extracted as described in Pasquet et al. (2016). Quantification of fungal DNA was realized by qPCR (see below) on 10 ng of total DNA using primers specific for the 18S ribosomal subunit-encoding genomic region (Mudge et al. (2006; Supplementary Table 2).

Complementary DNA synthesis was performed on 1 μg of total RNA using the ImProm-II^TM reverse transcription system (Promega France, Melun-les-Charbonnières, France) according to the manufacturer’s instructions. The resulting product was diluted 10 times in nuclease-free water. Primers were designed to amplify plant gene transcripts (Supplementary Table 2), including reference genes Bradi4g00660 (UBC18) and Bradi4g41850 (ACT3-like under accession number XM_003578821 in the nucleotide NCBI database) as previously determined by Hong et al. (2008); (Supplementary Table 2). qPCR reactions were performed on 2 μL of the diluted cDNA product using 8 pmoles of each specific primer and 10 μL of SYBRGreen Master Mix in a final volume of 20 μL. Reactions were performed in a Light Cycler LC480 Real-time PCR system (Roche Diagnostics, Meylan, France). All qPCR reactions were carried out on biological triplicates, each in technical duplicate. The final threshold cycle (Ct) values were the mean of three values (biological triplicates), each corresponding to the mean of technical duplicates. The comparative ΔΔCt method was used to evaluate the relative quantities of each amplified product in the samples. The Ct was automatically determined for each reaction by the Light Cycler LC480 Real-time PCR system set with default parameters. The specificity of the qPCR reactions was determined by melt curve analysis of the amplified products using the standard method installed in the system.

Detection and quantification of SLs exuded from the roots has been performed according to a method adapted from Ravazzolo et al. (2019). B. distachyon lines were grown hydroponically and exudation was performed in fresh 14 MS-KH₂PO₄ medium during 24 h. When necessary, exudates were immediately frozen in liquid nitrogen and conserved at −80°C before analysis. SLs were extracted from liquid medium complemented with 10 ng of (rac)-GR24 as internal standard, by adding the same volume of ethyl acetate followed by a manual and vigorous mixing during 10 min. Organic phase was decanted and dried using a rotary evaporator (Rotavapor, Buchi). The solid phase was resuspended in acetonitrile, and conserved at −20°C before analysis according to Boutet-Mercey et al. (2017). Separation was performed on a BEHC₁₈ column (2.1 × 100 mm, particle size 1.7 μm, Waters), using an ACQUITY UPLC I-class system (Waters), and detection on Waters Xevo TQ-S equipped with an ESI source and operated in positive ion mode. Identification and quantification were performed by the very sensitive MRM mode in LC-MS/MS (Supplementary Table 3), however, the signal in the sample was not sufficient to obtain a fullscan MS/MS spectrum. We first optimized all the conditions of MRM acquisition on the orobanchol standard in full scan methods (Supplementary Figure 3A), and chose the two main MRM transitions, then we analyzed the sample with this MRM methods on both transitions to guarantee the identification (Supplementary Figure 3B).

Amino acid sequences were recovered from The Cytochrome P450 Homepage (Nelson, 2009) or through reciprocal BLAST analysis for the Hordeum vulgare protein sequences. Phylogenetic analyses were inferred from a multiple sequence alignment generated by CLUSTAL W and by using the Maximum Likelihood method based on the JTT matrix-based model, conducted on MEGAX software (Kumar et al., 2018).

A previous study has identified Bradi1g75310 as a gene induced by a DON-producing F. graminearum [FgDON⁺, 6.93 fold log₂ compared to the mock condition 96 hours post inoculation (hpi)] while it does not respond to a mutant strain unable to produce the mycotoxin [FgDON^–, 0.77 fold log₂ compared to the mock condition (96 hpi); Pasquet et al., 2014]. In order to validate these transcriptomic data, reverse transcription-quantitative PCR (RT-qPCR) was conducted on the same RNA samples as those used for microarrays. Results showed good correlation between microarrays and RT-qPCR data as the Bradi1g75310 gene was strongly induced by the FgDON⁺ strain (15.4 fold log₂ at 96 hpi) and far less (6.2 fold log₂ at 96 hpi) by the FgDON^– strain (Figure 1A). To determine whether expression could also be induced by the mycotoxin itself, similar experiments were conducted on spikes point-inoculated with DON along kinetics from 0 to 48 hours post application (hpa). Expression of the Bradi1g75310 gene was strongly and rapidly induced by DON reaching a maximum of 10.74 induction fold log₂ 24 hpa (Figure 1B). Overall, these results confirm that the Bradi1g75310 gene expression is strongly induced by DON either produced by F. graminearum during infection or as a commercial molecule. The rapid and strong induction of expression following application of the mycotoxin alone reinforces the idea of a direct inductive action of the molecule.

The Bradi1g75310 gene is 2178 base pairs (bp) long, including 48 and 18 bp of 5′- and 3′-untranslated regions (UTR), respectively. The nucleotide sequence is constituted of 5 exons and 4 introns and the 1,572-bp coding sequence encodes a putative 523 amino acid CYP (Supplementary Figure 1), which has been classified as belonging to the CYP711A subfamily and named BdCYP711A29 (Nelson, 2009).² This notation will be therefore used throughout this work to refer both to the protein (BdCYP711A29) and the gene (BdCYP711A29). According to this database, BdCYP711A29 belongs to an oligomeric subfamily comprising 4 other members: BdCYP711A5 (Bradi3g08360), BdCYP711A6 (Bradi1g37730), BdCYP711A30 (Bradi4g08970), and BdCYP711A31 (Bradi4g09040). Sequences from A. thaliana and Oryza sativa CYP711A proteins (1 and 5 members, respectively) were used together with B. distachyon CYP711A protein sequences to perform multiple protein sequence alignment of the full-length proteins and phylogenetic analysis. The CYP711A1 sequence from the non-vascular plant species Selaginella moellendorffii has been used to root the tree (Figure 2 and Supplementary Table 4). BdCYP711A5, HvCYP711A5 and O. sativa OsCYP711A5, as well as BdCYP711A6, HvCYP711A6, and OsCYP711A6, constitute two specific clades. The other B. distachyon, H. vulgare, and O. sativa CYP711A proteins [BdCYP711A29, BdCYP711A30, BdCYP711A31, HvCYP711A29 (also named HvMAX1), HvCYP711A30, OsCYP711A2, OsCYP711A3, and OsCYP711A4] strongly grouped with A. thaliana MAX1 (At2g26510), and were distributed in two subclades. One subclade contains all rice proteins and the barley and B. distachyon CYP711A30 proteins and exhibits a strong species structuration. The BdCYP711A29 protein belongs to the other subclade together with the barley HvCYP711A29 and the BdCYP711A31 protein. The BdCYP711A29 protein exhibits 55.8% with the A. thaliana MAX1 protein involved in SL biosynthesis, converting CL into carlactonoic acid (CLA; Abe et al., 2014). BdCYP711A29 also shares between 46 and 58% identity with O. sativa CYP711A proteins (A2, A3, A4), all but one recently demonstrated to be involved at different steps of SL biosynthesis downstream of CL (Cardoso et al., 2014; Zhang et al., 2014).

In order to better characterize the B. distachyon CYP711A gene family, the relative expression of the five genes was examined by RT-qPCR in three week-old roots and leaves, and spikes of five week-old plants. Three different categories could be depicted based on the CYP711A gene expression pattern (Figure 3). The first one, with BdCYP711A5 gene as a single member, exhibits an overall low expression level with stronger expression in leaves and spikes (Figure 3A). BdCYP711A6 and BdCYP711A29 genes belong to the second category characterized similarly to the first one by a low expression level, and a significantly higher expression in roots and spikes (Figure 3B). The third category contains the BdCYP711A30 and BdCYP711A31 genes showing higher expression in roots only (Figure 3C). These results therefore indicate that the five BdCYP711A genes exhibit different spatial expression in the plant.

As mentioned before, previous transcriptomic data indicated that BdCYP711A29 is the only member of this oligogenic family to be differentially expressed following F. graminearum infection in a DON-dependent manner. To validate these data, RT-qPCR experiments using primers specific of the BdCYP711A5, BdCYP711A6, BdCYP711A30, and BdCYP711A31 genes (Supplementary Table 2) were conducted on the same samples as for the BdCYP711A29 gene. Following DON application, no significant induction of expression was detected for any of the four genes (Supplementary Figure 4A) apart from a decreased expression of the BdCYP711A6 gene from 12 hpa compared with the mock-inoculated control spikes. The expression of the BdCYP711A5, BdCYP711A30, and BdCYP711A31 genes was induced following infection by the F. graminearum FgDON+ strain, at different timepoints: [from 168 hpi onward for BdCYP711A5, 48 hpi for BdCYP711A30 and 96 hpi for BdCYP711A31 (Supplementary Figure 4B)]. However, besides the difference in its timing, the magnitude of induction is by no means equivalent to the one observed for the BdCYP711A29 gene (Figure 1). Indeed, whereas the maximal level of induction of BdCYP711A29 at 48 hpi reaches more than 15-fold log2 (that is more than 18,000 fold in absolute values, Figure 1A), maximal levels of expression for the BdCYP711A5, BdCYP711A30, and BdCYP711A31 correspond at most to a 5, 6.5, and 17-fold induction compared with the mock-inoculated control (Supplementary Figure 4B). From these results, we can conclude that, among the five BdCYP711A genes, BdCYP711A29 is the only gene copy which exhibits such a strong transcriptional induction following F. graminearum infection in a DON-dependent manner.

Following Agrobacterium tumefaciens-mediated transformation of the B. distachyon Bd21-3 ecotype (see section “Materials and Methods”), two independent homozygous BdCYP711A29 overexpressing lines were recovered (data not shown),. In addition, a null segregant, NS-11.26, that is a plant regenerated from the in vitro culture steps but devoid of the transgene, was selected to be used as a control (Table 1). RT-qPCR experiments were conducted to verify the expected overexpression of the BdCYP711A29 gene in spikes, leaves and roots. As shown in Figure 4, overexpression rates were highly similar in the three organs with line OE-CYP12.20 exhibiting a significantly higher level of expression than line OE-CYP11.29. These two overexpressing lines were used in further experiments (Table 1).

Lines mutated for the BdCYP711A29 gene were obtained by screening the B. distachyon TILLING mutant collection available in the Bd21-3 ecotype (Dalmais et al., 2013, see section “Materials and Methods” and Supplementary Figures 1, 2). Eleven mutant families carrying point mutations in the region of interest were recovered (Supplementary Table 5). A Sorting Intolerant From Tolerant (SIFT) analysis (Kumar et al., 2009) showed that four mutant families exhibited a SIFT score below 0.05 and were therefore likely to carry point mutations predicted to strongly impact the functionality of the protein (Supplementary Table 5): three families exhibited missense mutations (8,687, 7,708, and 7,424) and one (5,374) carried a nonsense mutation (R450*) leading to a protein truncated of the last 73 amino-acids, therefore lacking the highly conserved heme-binding domain (Supplementary Figure 2). Control lines carrying a BdCYP711A29 WT allele could be selected for families 8,687 and 5,374 but not for two other mutant families 7,424 and 7,708, which is problematic considering the numerous mutations per genome in TILLING mutant families. Therefore, only the two mutant lines, notated M8687#12 and M5374#135, respectively, and their corresponding controls, notated WT8687#2 and WT5374#139, respectively, were considered for further analyses (Table 1).

As previously mentioned BdCYP711A29 belongs to the CYP711A subfamily and is closely related to A. thaliana MAX1 and rice MAX1-like CYPs (Figure 2). Members of this subfamily are involved in SL biosynthesis (Bak et al., 2011; Stirnberg et al., 2002; Booker et al., 2005; Challis et al., 2013; Zhang et al., 2014). However, as for other multicopy families of proteins involved in plant secondary metabolism, phylogeny is usually insufficient to demonstrate functional homology (Moghe and Kruse, 2018). To determine whether the B. distachyon BdCYP711A29 is involved in SL biosynthesis, SL quantification was performed from root exudates.

First, we performed preliminary experiments on root exudates from the WT line collected after 1 week of phosphate starvation using ethyl acetate supplemented with 10 ng (rac)-GR24 as the internal standard (see section “Materials and Methods” and Supplementary Figure 3). Analyses were performed by LC-MS/MS via Multiple Reaction Monitoring (MRM). Supplementary Table 3 lists the chromatographic peaks of characteristic m/z MRM transitions detected at least in one biological replicate as compared to blank sample. Both orobanchol- and strigol-type SLs were found but most of detected transitions were not specific to one SL (e.g., 345 > 97, 345 > 248 and 367 > 270 could be linked to both didehydroorobanchol and didehydrostrigol, Supplementary Table 3). We therefore focused on compounds detected through at least two different characteristic transitions: orobanchol, orobanchyl acetate, solanacol, and solanacyl acetate, each giving a chromatographic peak at only one retention time (RT). Among these, only orobanchol characteristic transitions were recorded at the same RT (9.2 min). An additional experiment using the WT line allowed us to detect a chromatographic peak at the same RT (9.2 min) for 3 orobanchol characteristic transitions (347 > 205; 347 > 97; and 347 > 233). Finally, this RT was confirmed by adding orobanchol standard in samples, which resulted in a single peak at this same RT for each transition and similar ratios as in WT line (Supplementary Figure 3B). Therefore, following MRM, we were able to confirm the presence of orobanchol in B. distachyon WT root exudates.

Second, the relative amount of orobanchol exuded from the roots of three selected contrasting lines, the WT, overexpressing (OE-CYP12.20) and TILLING mutant (M5374#135) lines was then measured by calculating the ratio between the area under the curve (AUC) of the 347 > 97 chromatographic peak, specific to orobanchol, and the AUC of the 321 > 224 chromatographic peak, and comparing it to values of the internal standard (rac)-GR24 (Boutet-Mercey et al., 2017). Data were normalized over root fresh weight. We observed a significant increase of about 16 times of the relative quantity of orobanchol in OE-CYP12.20 exudates, as compared to WT and M5374#135 lines (Figure 5 and Supplementary Table 6). These results strongly suggest that the B. distachyon BdCYP711A29 protein is involved in orobanchol biosynthesis. No significant difference was observed between Bd21-3 and the TILLING mutant line, suggesting potential functional redundancy with one or several B. distachyon other CYP711A copies.

Considering that the B. distachyon BdCYP711A29 is involved in orobanchol biosynthesis, 4 weeks-old (end of the vegetative stage in our conditions) plants of overexpressing lines and TILLING mutants and their respective control lines were examined for the number of tillers and compared to the WT Bd21-3 ecotype. Apart from control and mutant lines WT5374#139 and M5374#135, no statistically significant difference could be observed compared to the WT line (Supplementary Figure 5). This indicates that BdCYP711A29 does not significantly participate to B.distachyon development.

In order to determine whether the BdCYP711A29 gene is involved in the interaction between B. distachyon and F. graminearum, spray inoculations by spore suspensions of the F. graminearum FgDON⁺ strain were performed on whole spikes of the lines previously described: WT, overexpressing lines (OE-CYP11.29 and OE-CYP12.20) and the corresponding null segregant (NS-11.26), TILLING mutant lines (M5374#135, M8687#12) and their control lines (WT5374#139 and WT8687#2, respectively). Representative symptoms, that is more or less extended bleaching on spikelets, at 7 and 14 dpi are presented in Figure 6A. FHB symptoms were quantified at 7 and 14 dpi by counting the number of spikelets exhibiting more than 50% of symptomatic florets over the total number of spikelets (Figure 6B). The results represent the mean of four independent biological replicates. Seven dpi, both overexpressing lines OE-CYP11.29 and OE-CYP12.20 exhibited more developed symptoms (35.9 ± 4.7% and 46.2 ± 6.1% of symptomatic spikelets, respectively) as compared to WT (11.2 ± 3.3%) and NS-11.26 (14.4 ± 3.3%) (Figure 6B). All the TILLING mutant lines, carrying either the mutant or WT allele for the BdCYP71A29 gene, presented symptoms equivalent to those observed with the WT line (Figure 6B). After another week of disease development (14 dpi), the tendency described above was conserved. Both overexpressing lines presented increased symptoms as compared with control lines (48.3 ± 4.9% for OE-CYP11.29 and 68.6 ± 5.2% for OE-CYP12.20 compared to 32.5 ± 5.1% and 26.3 ± 5.0% for Bd21-3 and NS-11.26, respectively) (Figure 6B). The approx. 20% differential development of symptoms between the two overexpressing lines was statistically significant (Figure 6B). As observed at 7 dpi, the TILLING lines did not show any significant difference in the level of symptomatic spikelets as compared to the WT line (Figure 6B).

To be able to correlate the differences of disease symptoms described above with a differential fungal development, we quantified by qPCR the relative amount of fungal gDNA on the same material used for symptoms quantification (3 out of 4 biological replicates). At 7 dpi, WT samples exhibited 20.3 ± 10.8% of fungal DNA and the other control lines showed similar values: 24.7 ± 11.3; 11.9 ± 7.6 and 21.6 ± 14.9% for NS-11.26, WT5374#139 and WT8687#2, respectively (Figure 7). Whereas TILLING mutant lines did not exhibit differences greater than 10% compared to WT and TILLING control lines (22.5 ± 11.6 and 13.4 ± 10.9% for M5374#135 and M8687#12, respectively), both overexpressing lines contained an increased percentage of fungal DNA as compared to Bd21-3 and NS-11.26 (45.8 ± 16.4 and 62.8 ± 12.5% for OE-CYP11.29 and OE-CYP12.20, respectively, Figure 7). This tendency was conserved at the 14 dpi time-point since all the control lines and TILLING mutant lines ranged from 28.7 ± 15.6 (M5374#135) to 40.6 ± 5.4% (M8687#12), contrary to overexpressing lines which exhibited from around 20–45% more fungal DNA (59.0 ± 4.3 and 75.5 ± 8.9% for OE-CYP11.29 and OE-CYP12.20, respectively, Figure 7). Statistical analyses were conducted on these results and showed that the OE-CYP lines contained significantly more fungal DNA compared to the control lines Bd21-3 and NS-11.26 either at one (OE-CYP11.29, 7 dpi) or both time-points (OE-CYP12.20, 7 and 14 dpi) therefore confirming the increased susceptibility of these lines to fungal infection (Figure 7).

To determine if the observed increased susceptibility of the overexpressing lines may be due to a modification of plant defenses, expression of B. distachyon defense marker genes previously shown to be induced upon F. graminearum infection (Pasquet et al., 2016) was investigated by RT-qPCR on spikes of the WT line, the NS-11.26 null segregant, the two overexpressing lines (OE-CYP11.29 and OE-CYP12.20), the two TILLING mutant lines (M5374#135 and M8687#12) and their corresponding controls (WT5374#139 and WT8687#2, respectively). To do so, we used four defense marker genes, two genes encode pathogenesis-related proteins, PR1-5 (Bradi1g57590, Kakei et al., 2015; Kouzai et al., 2016) and PR9 (Bradi1g39190, Pasquet et al., 2014), the third one encodes a phenylalanine ammonia lyase, recently renamed BdPAL6 (Bradi3g47110, Cass et al., 2015) and the last one codes for a uridine diphosphate (UDP)-glucosyltransferase shown to be involved in the conjugation of DON into DON-3-O-glucose (Bradi5g03300, Pasquet et al., 2016). The expression of these four defense genes was first assessed on healthy spikes. Very low basal levels of expression prevented proper comparison of the expression levels in healthy plants of the different genotypes (data not shown). Nevertheless, relative expression of the four marker genes could be quantified after spray inoculation by the F. graminearum FgDON⁺ strain. The 96 hpi time-point was chosen because it corresponds both to the infection time-point used in the transcriptomic analysis of B. distachyon response to fungal infection (Pasquet et al., 2014) and to maximal expression of most of the defense genes in our conditions (Schweiger et al., 2013; Pasquet et al., 2016). Almost no statistically significant difference could be detected between the different lines, whatever the alteration of the BdCYP711A29 gene, overexpression or mutation (Figure 8). The unique difference was observed for the PR9 gene (Bradi1g39190, Figure 8A), for which a stronger induction was observed 96 hpi in the OE-CYP12.20 line as compared to the WT Bd21-3 line (1.6 fold log₂) and to the NS-11.26 null segregant (1.7 fold log₂). These results suggest that the increased susceptibility of lines overexpressing the BdCYP711A29 gene likely does not result from an alteration of plant defense responses.

Strigolactones have been shown to promote pre-symbiotic growth of AMF (Akiyama et al., 2005) and more recently to regulate root entry through the modulation of hyphopodia formation (Kobae et al., 2018). As mentioned in the Introduction part, the picture is less clear for other fungi, in particular for plant pathogenic fungi (Steinkellner et al., 2007; Dor et al., 2011; Torres-vera et al., 2014; Foo et al., 2016; Belmondo et al., 2017; Decker et al., 2017). No major difference in plant defense gene expression could explain the increased susceptibility of lines overexpressing the BdCYP711A29 (Figure 8) and the OE-CYP12.20 line overexpressing the BdCYP711A29 gene exudates increased levels of orobanchol (Figure 5). We therefore investigated the direct impact of this SL on F. graminearum, and more specifically on the germination of typical 5–6 cell asexual spores named macroconidia (Booth, 1975). Various concentrations of orobanchol were tested and their impact on the germination of F. graminearum macroconidia was estimated using an in vitro test on small agar slides. As shown in Figure 9A (see also Supplementary Figure 6 for characteristic macroconidial germination patterns at each orobanchol concentration), orobanchol increased the number of fungal germ tubes per macroconidium after 12 h incubation with a peak at 10^–10 M. These results show that orobanchol stimulates F. graminearum macroconidial germination.

To further establish the relationship between alteration of the BdCYP711A29 gene or of its expression and F. graminearum macroconidial germination, similar experiments were conducted using root exudates of the WT, overexpressing (OE-CYP11.29 and OE-CYP12.20) or mutant (M5374#135 and M8687#12) lines obtained in phosphate-deprived conditions (see section “Materials and Methods” and Figure 9B) in place of pure orobanchol. After 12 h incubation at room temperature, germination percentages were 11.90 ± 2.83% if exudates were obtained from Bd21-3 (WT) roots (Figure 9B). The values were significantly increased if root exudates from the OE-CYP11.29 (24.90 ± 3.20%) or the OE-CYP12.20 (27.15 ± 2.23%) lines were used (Figure 9B). No significant differences were observed with root exudates from the mutant line M5374#135 (9.51 ± 2.30) nevertheless a significant difference (reduction) observed for the mutant line M8687#12 (7.05 ± 0.43) (Figure 9B). These results show that altering the BdCYP711A29 gene sequence or expression modifies the composition of root exudates and consequently their ability to promote F. graminearum macroconidial germination. Together with the data obtained with commercial orobanchol (Figure 9A) and the analysis of SL amounts shown previously (Figure 5), the modification of germination presumably occurs through a modulation of orobanchol content.

In this work, we have functionally characterized a CYP-encoding gene from B. distachyon, BdCYP711A29 (Bradi1g75310), previously identified as differentially induced in a transcriptomic study of the plant response to F. graminearum-produced DON during infection (Pasquet et al., 2014). Detailed analysis of the deduced amino-acid sequence (Supplementary Figure 1) revealed that the gene encodes the B. distachyon BdCYP711A29 protein of the CYP711A subfamily comprising four additional members in B. distachyon [BdCY711A5, A6, A30, and A31 (Figure 2)]. The BdCYP711A29 protein belongs to the same clade as the A. thaliana MAX1 protein involved in the conversion of CL into CLA (Abe et al., 2014) and as two O. sativa CYP711A proteins (OsCYP711A2 and A3) involved in SL biosynthesis downstream of CL (Cardoso et al., 2014; Zhang et al., 2014, Figure 2). We validated differential induction of BdCYP711A29 expression following infection by a DON-producing F. graminearum strain compared to a mutant strain unable to produce the mycotoxin and also showed that expression was induced following direct application of the mycotoxin (Figure 1). We further showed that the expression of the four other BdCYP711A genes were either not or far less regulated in the same conditions (Supplementary Figure 4). The five members of the oligogenic family were differentially expressed in B. distachyon organs with BdCYP711A5 being more expressed in leaves and spikes, BdCYP711A6 and A29 more expressed in roots and spikes whereas BdCYP711A30 and A31 are almost exclusively expressed in roots (Figure 3). These results suggested that, based on the expression of the five corresponding genes, the proteins may either play redundant roles (BdCYP711A6 and A29 or BdCYP711A30 and A31) or differential ones along the whole plant.

To investigate the role of BdCYP711A29 in SL biosynthesis, B. distachyon lines altered in BdCYP711A29 gene expression or sequence were generated or selected. Two transgenic lines carrying a single T-DNA insertion and overexpressing the gene were generated as well as their corresponding null segregant (Figure 4 and Table 1). Using a TILLING mutant collection available in our laboratory (Dalmais et al., 2013), two mutant lines were selected both carrying mutations predicted to strongly impact the CYP functionality, together with their control lines carrying a WT BdCYP711A29 allele (Table 1 and Supplementary Figure 2 and Table 4). Based on the role of SLs in plant development (Al-Babili and Bouwmeester, 2015), OE and mutant lines were characterized for shoot branching by measuring the tillers number and compared to the WT and their corresponding control lines. No statistically supported difference related to the alteration of the BdCYP711A29 sequence or expression could be observed (Supplementary Figure 5) suggesting that BdCYP711A29 may not play per se a significant role in B. distachyon shoot development. Alternatively, functional redundancy with one of the four other CYP711A copies encoded by the B. distachyon genome could occur considering the expression patterns of the five genes (Figure 3).

Due to their very low amounts in plant tissues, SLs are difficult to quantify in planta. Nevertheless, when plants are grown on phosphate-deprived media, root exudation is stimulated allowing their detection and quantification (López-Ráez et al., 2008). To determine whether BdCYP711A29 is involved in SL biosynthesis, root exudates of the strongest overexpressing line OE-CYP12.20, the mutant line carrying a STOP codon M5374#135 and the WT line were extracted and relative quantification of SLs was performed. A 16-fold increase in orobanchol was observed in the OE line as compared to the WT line (Figure 5). Few studies have deeply investigated the role of MAX1-like proteins in SL biosynthesis and diversification. In rice, as in B. distachyon, five CYP711A proteins are found, all but one (Os1500) being functional (Zhang et al., 2014). Using in vitro co-expression studies, Zhang et al. (2014) have shown that among the four functional rice MAX1-like proteins, Os900 is a CL oxidase, catalyzing the conversion of CL into ent-22-epi-5-deoxystrigol whereas Os1400 catalyzes the conversion of the latter product into orobanchol. The BdCYP711A29 protein groups in the same clade as Os900 and Os1400, together with two other B. distachyon proteins, BdCYP711A30 and BdCYP711A31 (Figure 2). Altogether, our results strongly suggest that the BdCYP711A29 is involved in orobanchol biosynthesis. Additional experiments involving in vitro co-expression studies in yeast or in Nicotiana benthamiana leaves as described in Zhang et al. (2014) will determine the exact step catalyzed by the BdCYP711A29 protein.

Few CYPs have been shown to be involved in the interaction between cereals and F. graminearum and only two of them have been characterized as strongly associated with resistance to DON. Ito et al. (2013) identified a CYP from the Gram-negative bacterium Sphingomonas sp. able to convert DON into 16-hydroxy-deoxynivalenol. More recently, virus-induced gene silencing of wheat TaCYP72A copies demonstrated their involvement in resistance to DON and grain development (Gunupuru et al., 2018). Among the five B. distachyon CYP711A-encoding genes, BdCYP711A29 is the only one strongly induced in spikes in a DON-dependent manner (Figure 1 and Supplementary Figure 4), suggesting a specific role in FHB. To assess the involvement of BdCYP711A29 in FHB, spikes of the different lines were spray-inoculated by a DON-producing F. graminearum strain and symptoms were scored 7 and 14 days after inoculation. Whereas no difference was observed between the WT ecotype Bd21-3, the null segregant NS-11-26 and the TILLING mutant lines or their corresponding control lines, the two OE lines, OE-CYP11.29 and OE-CYP12.20 were significantly more susceptible to FHB whatever the time-point (Figure 6). These results were confirmed by the quantification of fungal biomass by qPCR (Figure 7).

Strigolactones have been shown to be involved in complex cross-talks with other phytohormones (recently reviewed in Omoarelojie et al., 2019). In particular, tomato SL biosynthesis mutants were shown to have reduced JA content (Torres-vera et al., 2014). However, a more recent study in Arabidopsis has revealed no clear relationship between SLs and JA (Rozpadek et al., 2018). In contrast, in the same study, the authors have provided evidence of a complex crosstalk between SLs and SA: high doses of GR24 (1 μM) strongly induced SA biosynthesis in WT plantlets and max1 mutants exhibited increased SA content (Rozpadek et al., 2018). Nevertheless, reports are not sufficient yet to fully decipher the relationships between SLs and other phytohormones in plant-biotic interactions. In our study, although no SA or JA quantification has been performed in the different lines, we could show that increased susceptibility is unlikely correlated to an alteration of plant defenses. Indeed, no major difference in the induction of defense marker genes frequently used in our team (Pasquet et al., 2014), could be observed (Figure 8). Even if we cannot rule out that some specific defense pathways may be altered in some lines, major changes in plant defense is therefore not the explanation for increased susceptibility of the OE lines.

Beyond their role in planta, SLs have been shown to have a direct impact on the development of fungal microorganisms (De Cuyper and Goormachtig, 2017; López-Ráez et al., 2017). In vitro studies using the synthetic analog GR24 provided contrasted results likely due to variation in experimental set ups rather than to species-specific differences. As an example, the impact of GR24 on the growth of the fungal ascomycete B. cinerea has been investigated in several studies but has led to contradictory results (Steinkellner et al., 2007; Dor et al., 2011; Torres-vera et al., 2014; Belmondo et al., 2017). Given the increased exudation of orobanchol by the OE line OE-CYP12.20, we have investigated the impact of a range of physiological concentrations of this SL on F. graminearum. We have shown that orobanchol concentrations from 10^–12 to 10^–6 M significantly stimulated germination of F. graminearum macroconidia as exemplified by an increased number of germ tubes (Figure 9A, Supplementary Figure 5). Using exudates recovered from phosphate-deprived hydroponic cultures of WT, BdCYP711A29 overexpressing or mutant lines, we have further shown that F. graminearum macroconidial germination was significantly increased in root exudates of the two overexpressing lines (OE-CYP11.29 and OE-CYP12.20) whereas it was reduced in exudates of one of the mutant lines (M8687#12) compared to the WT line Bd21-3 (Figure 9B). Together with the increased orobanchol content found by LC-MS/MS in root exudates of the OE-CYP12.20 line, these results strongly support a role of this SL in the stimulation of germination of F. graminearum macroconidia. Such a role is highly reminiscent of the well-documented role of SLs in the stimulation of spore germination of AMF (García-Garrido et al., 2009). F. graminearum beyond its ability to infect cereal spikes, has been also shown to be a root pathogen (Wang et al., 2015). The impact of plant root exudates and more specifically of orobanchol on the stimulation of spore germination of this soil-borne fungus is therefore biologically relevant. In a recent study, Ding et al. (2020) have set up an experimental design to quantitatively assess F. graminearum root infection using B. distachyon as a model plant. We were however unable to reproduce this experimental set up and future experiments are needed to better correlate increased orobanchol in root exudates with F. graminearum aggressiveness on plant roots.

Considering spike infection, SL quantification in spikes of the different B distachyon lines has not been performed in this study due to the very low concentrations of SLs in planta. Nevertheless, the expression pattern of BdCYP711A29 indicates both a higher basal expression in spikes as well as a strong induction of expression in the same organ at early steps of F. graminearum infection in a DON-dependent manner (Figures 1, 3). These results are therefore consistent with an involvement of this gene in the plant response during infection, likely through increasing orobanchol contents.

Altogether, the results obtained in this study strongly suggest that during the early steps F. graminearum infection the expression of the B. distachyon BdCYP711A29 gene is induced which increases orobanchol production in its targeted organ. As a stimulant of spore germination, orobanchol may consequently promote infection and FHB development.