A17 and F83005.5 M. truncatula accessions were used in each A. euteiches inoculation experiment as resistant and susceptible lines, respectively (Djébali et al., 2009). The nf-ya-1-1 mutant was obtained from A17 EMS mutagenized seeds. A glutamine to stop codon substitution in position 137 out of 322 amino acids leads to non-sense mutation coding for a truncated non-functional protein depleted of DNA and protein-protein interactions domains (Laporte et al., 2014). Plants were grown in vitro with a 16-h light at 22°C and 8-h dark, 20°C, as previously described (Djébali et al., 2009). For root transformation, we used the ARqua1 strain of Agrobacterium rhizogenes as described by Boisson-Dernier et al. (2001, 2005) to produce composite plants. Composite plants were grown in the same condition than seedlings, as described in Djébali et al. (2009). Root organ cultures were produced using the same transformation pipeline and then propagated in absence of aerial parts on media supplemented with sucrose as described in Genre et al. (2013).

Zoospores of A. euteiches, a pea isolate, were produced as described by Badreddine et al. (2008) and inoculated according to methods described in Rey et al. (2013). Four independent in vitro infection assays were performed with 15 to 30 inoculated plants per line in each repeat. The calculated means for each parameter were compared through statistical ANOVA analyses.

For optical microscopy, 100 μm root sections were prepared from fresh infected seedlings and labeled with WGA-FITC, as described by Djébali et al. (2009) and Rey et al. (2013) to localize A. euteiches hyphae using epifluorescence illumination (excitation filter, BP 450–490 nm). Images were acquired using a CCD camera (color Coolview, Photonic Science, Robertsbridge, UK). Image J¹ was used to perform hyphal counts of A. euteiches in root cross sections. Following image acquisition, the green channel of the pictures (which shows WGA-FITC signal) was transformed in a binary image where hyphae appear in black and then subjected to a particle analysis with standard settings.

The miR169q overexpression was conducted with constructs designed by Combier et al. (2006). Briefly, a fragment of the pre-miR169q located just outside of the stem-loop structure was amplified and then cloned into pPex (Combier et al., 2006). Entry clones for the GUS open reading frame (Control) and the 3′UTR of MtNF-YA1 were obtained in the Gateway vectors pDONR207 and pDONRP2R-P3, respectively, and recombined in pPEX-GUS. For pMtNF-YA1 promoter analysis in M. truncatula, entry clones were recombined in the binary vector pK7m34GW (Laporte et al., 2014).

Total RNAs were isolated from inoculated and non-inoculated roots with the “RNeasy for plant and fungi” kit (Qiagen). Following total RNA extraction and DNAse treatment, RNA integrity was checked using a Bioanalyzer (Agilent technologies) and reverse transcriptions were performed with 1 μg of total RNA for each sample using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) with random primers in a total volume of 20 μL. Gene specific primers were designed using the Quantprime software² (Arvidsson et al., 2008). Then, the cDNAs were used for high-throughput qPCR using the BioMark^TM HD System (Fluidigm): first, 1.3 μL of a 1:40 dilution of the synthesized cDNA were submitted to specific target amplification (STA) by PCR amplification in a 5 μL reaction containing 96 M. truncatula specific primer pairs (50 nM each) and a twofold dilution of the TaqMan^® PreAmp Master Mix (Applied Biosystems). The PCR program consisted of 14 cycles of 15 s at 95°C followed by 4 min at 60°C. Then, 340 nL of preamplified cDNA were used for qPCR array analysis in a 6.7 μL reaction using EvaGreen chemistry (Applied Biosystems). Data were analyzed with the BioMark Real-Time PCR Analysis Software Version 2.0 (Fluidigm) (Vergnes et al., 2014). Calculations for comparing expression data were performed using the 2^-ΔCt and the 2^-ΔΔCt methods (Livak and Schmittgen, 2001) using means Ct values of three M. truncatula reference genes which were selected using the NormFinder software (Andersen et al., 2004): Histone-3-like (Medtr4g097170), Ubiquitin family protein (Medtr3g097170) and Translation elongation factor 1 (Medtr6g021800) were used to standardize expression in seedlings. Medtr4g097170 and Medtr6g021800 were used to normalize expression in composite plants whilst only Histone-3-like was analyzed in root organ culture experiments due to variable expression profile of the other housekeeping genes in these two types of biological material (Supplementary Figure 1). Gene identifier, primers, amplicons length and melting temperature are listed in (Supplementary Table 1). Gene encoding an A. euteiches tubuline transcript was used to assess the amount of pathogen biomass in planta at 6 dpi. For detection of miR169q overexpression, primers designed just outside the stem-loop of the pre-miR were used to detect transcription in plant cDNA.

Total RNA was extracted using the RNeasy Kit (Qiagen). One microgram of total RNA was used to produce labeled cRNA and hybridize Affymetrix GeneChip^® M. truncatula genome arrays at INRA-URGV (Evry, France). Three independent biological replicates were performed for A17 wild-type (WT) and Mtnf-ya1-1 mutants, non-inoculated or harvested 1 and 6 days post inoculation (dpi). For each biological repetition RNA samples were extracted from ten roots of 15-day-old plants. Following extraction, RNA samples were checked for their integrity on the Agilent 2100 bioanalyzer according to the Agilent Technologies (Waldbroon, Germany). One microgram of total RNA was used to synthesize biotin-labeled cRNAs with the One-cycle cDNA synthesis kit (Affymetrix, Santa Clara, CA, USA). Superscript II reverse transcriptase and T7-oligo (dT) primers were used to synthesize the single strand of cDNA at 42°C during 1 h followed by the synthesis of the double stranded cDNA by using DNA ligase, DNA polymerase I and RNaseH during 2 h at 16°C. Clean-up of the double-stranded cDNA was performed with Sample Cleanup Module (Affymetrix) followed by in vitro transcription (IVT) in presence of biotin-labeled UTP using GeneChip^® IVT labeling Kit (Affymetrix, Santa Clara, CA, USA). Quantity of the labeled-cRNA with RiboGreen^® RNA Quantification Reagent (Turner Biosystems, Sunnyvale, CA, USA) was determined after cleanup by the Sample Cleanup Module (Affymetrix). Fragmentation of 15 μg of labeled-cRNA was carried out for 35 min at 94°C, followed by hybridization during 16 h at 45°C to Affymetrix GeneChip^® M. truncatula Genome Array representing approximately 61,200 probes: 32,167 M. truncatula EST/mRNA-based probe sets; 18,733 M. truncatula IMGAG and phase 2/3 BAC prediction-based probes; 1,896 M. sativa EST/mRNA based probes; and 8,305 S. meliloti gene prediction-based probes. After hybridization, the arrays were washed with two different buffers (stringent: 6X SSPE, 0.01% Tween-20 and non-stringent: 100 mM MES, 0.1M [Na+], 0.01% Tween-20) and stained with a complex solution including Streptavidin R-Phycoerythrin conjugate (Invitrogen/molecular probes, Carlsbad, CA, USA) and anti-Streptavidin biotinylated antibody (Vectors laboratories, Burlingame, CA, USA). The washing and staining steps were performed in a GeneChip^® Fluidics Station 450 (Affymetrix). The Affymetrix GeneChip^® M. truncatula Genome Arrays were finally scanned with the GeneChip^® Scanner 3000 7G piloted by the Command Console Launcher Tool. The raw CEL files were imported in R software for data analysis³. The data were normalized with the gcrma algorithm, available in the Bioconductor package (Gentleman et al., 2004). Raw and normalized data are available through the CATdb database (AFFY_aphanomyces_M. truncatula) (Gagnot et al., 2008) and from the Gene Expression Omnibus (GEO) repository at the National Centre for Biotechnology Information (NCBI) (Turnpenny, 2008), A17 control and 1 day post inoculation data are available in GSE20587 and A17 6 dpi as well as nf-ya1-1 control, 1 and 6 dpi are available in GSE26046.

Genes regulated depending on genotype and time post inoculation were determined by ANOVA according to Moreau et al. (2013). To determine differential gene expression subsequent to ANOVA, we performed a usual two group t-test that assumes equal variance between groups. The variance of the gene expression per group is a homoscedastic variance, where genes displaying extreme variance (p < 0.001) were excluded. The raw p-values were adjusted by the Bonferroni method, which controls the Family Wise Error Rate (FWER). A gene was declared differentially expressed if its Bonferroni p-value was lower than 0.001. Multi experiment Viewer (MeV⁴) was used for hierarchical clustering of gene expression profiles using euclidean distance and average linkage parameters (Eisen et al., 1998). Annotations used in this study are based on the work of published by Czaja et al. (2012) and Roux et al. (2014). Genes were also classified according to MapMan mapping Mt3.0_AFFY_0510⁵. Venn diagram was performed on Mapman tool.

We used the Mtnf-ya1-1 null mutant line obtained in the WT A17 genetic background (Laporte et al., 2014), to assess the role of MtNF-YA-1 in the colonization by A. euteiches zoospores. Young seedlings of this Mtnf-ya1-1 mutant line were inoculated by A. euteiches zoospores, along with A17 and F83005.5 seedlings which served as tolerant and susceptible lines, respectively. Phenotypes of both lines upon inoculation are shown in Figure 1A. At 14 dpi, most F83005.5 plants were severely attacked and their development impaired when compared to uninfected seedlings. A17 plants had similar size as control but a browning reflecting colonization of the whole roots cortical tissues was observed. Inoculated Mtnf-ya1-1 plants displayed symptoms too, but parts of their primary roots as well as some newly formed secondary roots remained symptomless. The extent of rot symptoms and percentages of dead plants (Figure 1B) were measured for 150 seedlings of each genotype and both parameters support a significant decrease of symptomatic tissues and of death rate for Mtnf-ya1-1 compared to A17 (p-value <0.001). We also assessed putative changes in root architecture of each line in control conditions and upon invasion by A. euteiches. The number and length of secondary roots was similar between the three lines in control conditions (n = 30 plants). In contrast, drastic modification of the root architecture was observed for all lines; 3 weeks post inoculation (Figure 1C). In F83005.5, a strong decrease in length and number of roots after inoculation was observed whereas A17 showed a two fold increase in the LR numbers with an average of their total length similar to control plants. In Mtnf-ya1-1, number of LRs (+29% – p-value <0.05) and their length (+37% – p-value <0.001) were both significantly increased compared to A17 WT. Taken together, these results suggest that (i) root branching and development are positively correlated to resistance to A. euteiches and (ii) MtNF-YA-1 is a negative regulator of these responses in A17.

As Mtnf-ya1-1 plants showed reduced visual symptoms when challenged with A. euteiches, we decided to thoroughly observe the colonization process of the oomycete. To dissect the effect of MtNF-YA1 knock out on the colonization behavior of the pathogen, its mycelium was stained using Wheat Germ Agglutinin-FITC (Badreddine et al., 2008). By this mean, we monitored the dynamics of the colonization from initial penetration events to complete invasion using epifluorescence microscopy. At 3 dpi, the rhizodermis of F83005.5 was completely surrounded by hyphae, while colonization was sparser in A17 and Mtnf-ya1-1 (Figure 2A). Following initial superficial colonization, the mycelium crosses the epidermis and enters cortical tissues through wounds and intercellular spaces. Root cross sections were performed 6 days after inoculation to track pathogen progress. An extensive colonization of F83005.5 root tissues was recorded, notably within the inner cortex and vascular tissues (Figure 2B). Hyphae were spreading extracellularly from the root tip inoculation point, by following intercellular spaces. In A17, a similar behavior was observed with the exception that the central cylinder was not invaded. In comparison, the Mtnf-ya1-1 roots harbored a reduced amount of hyphae mainly restricted to the outer cortex compared to A17 (Figure 2B). Twenty-one days post inoculation, root tissues were heavily colonized in F83005.5 and numerous oospores were produced inside dead cells, indicating that the pathogen fulfilled its life cycle (Figure 2C). In A17, numerous intercellular hyphae invaded the entire cortex but never entered the vascular tissues and only a few oospores were detected, in line with previous studies (Supplementary Figure 3A) (Djébali et al., 2009, 2011). The Mtnf-ya1-1 mutant displayed consistently a reduced hyphal colonization mostly located in the outer root cortex whilst reduced colonization of the inner cortex was observed (Supplementary Figure 3B). Aphanomyces hyphae were excluded from the vascular system and almost no oospores were detected (Figure 2B; Supplementary Figures 3A,B). Taken together, these observations indicate that intra-radical colonization by A. euteiches along with its biological cycle were hampered in Mtnf-ya1-1.

To strengthen the role of MtNF-YA1 during the accommodation of A. euteiches, its expression was analyzed by RT-qPCR in F83005.5, A17 and Mtnf-ya1-1 at 0, 1, 3, and 6 dpi (Figure 3). We investigated the gene expression in the three lines grown in mock conditions and noticed a slightly stronger expression in F83005.5 compared to A17 albeit the difference was not significant. We then analyzed MtNF-YA1 expression patterns at 1, 3, and 6 days following inoculation. We observed a late induction of MtNF-YA1 expression in F83005.5 which peaked 6 days after inoculation at a fourfold level compared to non-inoculated samples. We assessed significance of this variation within F83005.5 and detected a p-value <0.05 between 6 dpi expression values and all the other time points of the kinetic in this genotype. In A17, MtNF-YA1 expression was low and stable throughout the kinetic. We compared gene expression between genotypes at 6 dpi and observed that A17 expression significantly differed from F83005.5 and was reduced to 13% of the expression level recorded in this susceptible genotype. The MtNF-YA1 expression observed in the nf-ya1-1 mutant was significantly lower at 1 day post inoculation than in A17 but otherwise similar to the barely detectable levels of transcripts detected in A17. The Mtnf-ya1-1 mutant produces an early stop codon transcript and Western blot analysis showed previously that the truncated protein was absent (Laporte et al., 2014). The very low level of MtNF-YA1 transcripts measured at 1 dpi as compared to A17 (Figure 3) further supports the hypothesis that this might be a consequence of non-sense mediated mRNA decay (NMD) (Kervestin and Jacobson, 2012) and that nf-ya1-1 does not express functional MtNF-YA1 transcripts.

We then checked if allelic variation in the promoter sequences of MtNF-YA1 between susceptible and resistant lines may explain the observed differential expression. We observed polymorphism both in the coding and promoter region of MtNF-YA1 by investigating SNP data from the M. truncatula HapMap Project⁶, on a set of sequences from 226 M. truncatula accessions (see Supplementary Data Sheet S1; Supplementary Figure 2). More precisely, the genetic distance (Kimura 2-parameters) between A17 and F83005.5 for both regions, and even for the promoter region only (∼2 kb), was among the top 3% highest pairwise distance values calculated, based on a total of 25651 pairwise comparisons (Supplementary Figure 3; Supplementary Table 5). This finding thus supports the hypothesis that the differential expression of MtNF-YA1 between A17 and F83005.5 could be due to increased divergence in the promoter region.

To validate the involvement of MtNF-YA1 in disease susceptibility, we decided to suppress its expression in the susceptible genotype F83005.5. First, the miR169q that negatively regulates the MtNF-YA1 transcript levels (Combier et al., 2006) was overexpressed in F83005.5 root organ cultures by Agrobacterium rhizogenes mediated transformation. To do so, a 35S::miR169q construct overexpressing the precursor of the miR (pre-mir169q) was transformed in M. truncatula roots. Their susceptibility level to A. euteiches was compared to root organ cultures of the same line transformed with an empty vector control by assessing the amount of pathogen through RT-qPCR. Three independent hairy root lines for each construct were selected on the basis of their growth speed for further work. Both control and miR169q overexpressing root organ cultures were inoculated with A. euteiches and harvested at 6 dpi for subsequent analysis. Using RT-qPCR, we checked the overexpression of pre-miR169q in transgenic lines and detected on average a 305-fold increase in pre-miR169q transcript levels as compared to root organ cultures transformed with control vector (p-value <0.001) (Figure 4A). Consequently, MtNF-YA1 transcript levels were reduced to 17% of the WT situation (p-value <0.001) (Figure 4A). Using the same samples, we quantified A. euteiches tubulin transcripts 6 days after root inoculation by using a RT-qPCR method as a proxy (Rey et al., 2013). This time point was chosen as it allowed detection of both pathogen and plant cDNAs and matches the timing of MtNF-YA1 transcriptional induction in F83005.5. Strikingly, a strong reduction of 88% in the A. euteiches tubulin transcripts was observed for 35S::miR169q root organ cultures compared to control root organ cultures (Figure 4A).

In a complementary approach, composite F83005.5 plants were transformed with control vector and an RNAi construct targeting specifically the 3′UTR of MtNF-YA1 (35S::aSIII’UTR). One hundred and twenty transformants were obtained for each construct and subsequently inoculated with A. euteiches. Six days after inoculation, transgenic root systems were pooled in 12 samples for each constructs. The RNAi strategy led to a significant decrease of the MtNF-YA1 transcript accumulation (63% lower than in controls). Overall colonization level of roots isolated from composite plants was lower than in the root organ culture experiments but the A. euteiches tubulin transcripts were also reduced in NF-YA1 silenced roots to 43% of their level in control roots (Figure 4B).

In conclusion, both miRNA and RNAi-based strategies led to a significant down-regulation of MtNF-YA1 expression and this correlated with an increased resistance of the susceptible genotype F83005.5 to A. euteiches.

To identify the gene network under control of the MtNF-YA1 transcription factor in interaction with A. euteiches, we carried out transcriptome analyses using whole genome Affymetrix chips on A17 and Mtnf-ya1-1. RNAs extracted from whole root systems were harvested in control conditions and one or 6 dpi. An ANOVA analysis was performed to detect differential gene expression at each time point among the two genotypes (“Genotype effect“) and between controls and inoculated conditions (“Inoculation effect“). For each regulated gene (Bonferroni-corrected p-value <0.001), the factors conferring the changes in gene expression were determined and assigned to genotype or inoculation factors, for more details regarding this analysis pipeline, see Moreau et al. (2013). We then sorted genes displaying at least a twofold induction or repression compared to control (fold >1 or <-1 in log2). Five thousand three hundred and eighteen probes corresponding to 3970 individual genes were found to be regulated by one or both factors mentioned above, representing almost 10% of the overall Affymetrix chip probes (50900 probes on the chip) (Supplementary Table 4). We validated by RT-qPCR a selection of regulated genes of both genotype and inoculation effects and obtained an overall good correlation in the fold change obtained by the microarray and the qPCR (R² = 0.78) (Supplementary Figure 1; Supplementary Table 1). In order to specifically assess the contribution of MtNF-YA1 to these responses, we focused on the 1965 probes representing 1509 genes affected by Genotype effect as they display differential expression between A17 and the Mtnf-ya1-1 mutant (Supplementary Tables 2 and 4). We produced a Venn diagram of probes overexpressed in Mtnf-ya1-1 or A17 using ratios between mutant and WT expression levels at each time point (threshold = fold 2 between genotypes). A vast majority of probes were altered in the ratio between control 0 dpi mutant and wild type plants, (grayed circle) indicating major basal differences between the root transcriptome of Mtnf-ya1-1 and A17 (Figure 5A). A hierarchical clustering of these 1965 probes revealed two patterns of gene regulation upon A. euteiches colonization either mostly induced (529 probes, 360 genes) or preferentially repressed (1436 probes, 1150 genes) (Figure 5B). Interestingly, expression of these 1965 probes in non-inoculated Mtnf-ya1-1 was similar to the gene expression observed in A17 at 1 and 6 dpi. A similar observation can be drawn from genes constitutively repressed in the mutant (Supplementary Tables 3 and 4). This last finding implies that part of the A17 transcriptomic responses to A. euteiches is already constitutively expressed in Mtnf-ya1-1 in non-challenged conditions. Hence, NF-YA1 appears as a central regulator of responses triggered in WT roots by A. euteiches invasion.

We looked closer into the biological functions of genes deregulated in Mtnf-ya1-1. Among the 529 probes induced in non-inoculated mutant control and infected A17, 36 genes are unambiguously related to stress responses already known to contribute to resistance toward A. euteiches (Supplementary Tables 2 and 4) (Colditz et al., 2004, 2005, 2007; Djébali et al., 2009, 2011; Schenkluhn et al., 2010; Kiirika et al., 2012; Rey et al., 2013). The 1436 probes showing repression spanned more diverse processes (Supplementary Tables 3 and 4). Among these, 20 genes involved in hormonal biosynthesis and signaling, notably auxins, ethylene and abscisic acid (ABA) are down regulated in Mtnf-ya1-1 already before inoculation and sometimes 1 and 6 dpi(Supplementary Table 3). These hormonal imbalances in the mutant may explain stress responses mentioned above or alternatively the modifications of root architecture observed upon invasion (De Smet, 2012; De Lucas and Brady, 2013; Tian et al., 2014).

Our transcriptomic approach revealed that 36 genes involved in resistance and stress mechanisms are constitutively expressed in the uninoculated Mtnf-ya1-1 mutant, thereby indicating MtNF-YA1 negatively regulates the expression of defense and stress-related genes in roots. The limitation of A. euteiches hyphal colonization in the root cortex observed after the knockout of MtNF-YA1 in the Mtnf-ya1-1 mutant as well as the knock down of this gene in the partially resistant A17 or the very sensitive F83005.5 could thus be explained by the constitutive expression of these defense and stress-related genes. Interestingly Laporte et al. (2014) also observed a negative effect of the knockout of MtNF-YA1 in the Mtnf-ya1-1 mutant on epidermal and cortical root colonization but by the symbiotic bacterium Sinorhizobium meliloti. This suggests that, despite the clear differences between root colonization by a pathogenic oomycete and by a beneficial bacterium some common mechanisms regulated by MtNF-YA1 may exist. Rhizobial infection strongly up-regulates MtNF-YA1 expression, especially in the tissues surrounding infection threads (Laporte et al., 2014). In the Mtnf-ya1-1 mutant infection progression is severely disturbed, possibly because MtNF-YA1 does not downregulate defense mechanisms as in A17 WT. It is possible that Aphanomyces euteiches has evolved to exploit this potential symbiotic mechanism and thus downregulates defense mechanisms by up-regulating MtNF-YA1 in susceptible ecotypes. How widespread this susceptibility mechanism is in the interaction with root filamentous pathogens or other root pathogens still has to be addressed.

In Arabidopsis thaliana, overexpression of the closest homolog of MtNF-YA1, AtNF-YA2 leads to a significant increase in LR primordia density (Sorin et al., 2014) and the authors propose that AtNF-YA2 controls root architecture. Furthermore Laporte et al. (2014) reported that MtNF-YA1 is also expressed transiently in LR primordia during LR initiation suggesting that this gene may also control LR growth in M. truncatula.

The resistance of M. truncatula roots to A. euteiches infection is correlated with the ability of the plant to generate new roots upon invasion by the pathogen (Djébali et al., 2009; Laporte et al., 2014; Laffont et al., 2015) (and this study). Here, we show that both the number and the length of LRs that the plant produces in response to an inoculation by A. euteiches are higher in the Mtnf-ya1-1 mutant compared to the A17 accession and higher in the A17 as compared to the F83005.5. There is thus a negative correlation between the level of expression of MtNF-YA1 and the capacity of the plant to initiate and elongate LRs upon inoculation.

Interestingly, the cytokinin receptor mutant cre1 mutant is also more resistant to A. euteiches infection and displays more LRs upon infection, suggesting that this might be a cytokinin controlled mechanism (Laffont et al., 2015). Our transcriptome analysis showed that a certain number of cytokinin-related genes are mis-regulated in Mtnf-ya1-1. Among them 2 genes coding for cytokinin-O-glucosyltransferases (ZOG), i.e., Medtr6g014660 and Medtr3g111140.1 and a gene coding for a cytokinin oxidase, i.e., Medtr4g118430.1 are significantly downregulated in Mtnf-ya1-1 roots compared to WT (A17) roots, both before inoculation and one day after inoculation. In Rice the suppression of ZOGs leads to a decrease in fasciculated root production (Kudo et al., 2012) while contradicting reports about the positive and negative effect of cytokinin oxidase on root growth exist (Köllmer et al., 2014; Zalewski et al., 2014).

Importantly, ABA signaling is known to render M. truncatula susceptible to A. euteiches (Colditz et al., 2005; Schenkluhn et al., 2010). Among the genes down regulated in the Mtnf-ya1-1 mutant compared to A17 both at day 0 and 1 day post inoculation, 6 encode ABA-related genes (Supplementary Table 3). Interestingly among them Medtr1g010210.1 encodes LATERAL DEFICIENCY-LATD. This nitrate transporter that could also transport ABA is required for nodule and LRs formation, (Bright et al., 2005; Liang et al., 2007; Harris and Dickstein, 2010; Bagchi et al., 2012).

Two auxin responsive genes, Medtr1g063950, encoding a SAUR protein and Medtr5g016320.1 encoding an indole-3-acetic acid-amido synthetase are also mis-regulated in Mtnfya1-1 (Supplementary Table 3). However, these two genes are only differentially regulated at day 0, prior to inoculation and only Medtr1g063950 remained overexpressed in the mutant one day post inoculation whilst Medtr5g016320.1 is less expressed in mutant at 6 dpi.

Taken together we have shown correlations between the expression of genes from hormonal pathways potentially involved in LR development and the MtNF-YA1 transcription factor that need to be further studied to understand the role of this gene in root development upon A. euteiches invasion. Unlike the situation found during the nitrogen fixing symbiosis where NF-YB (Soyano et al., 2013) or mycorrhization where NF-YC (Hogekamp et al., 2011; Gaude et al., 2012; Hogekamp and Küster, 2013) genes have been described as up-regulated, we did not find any mis-regulated NF-YB and NF-YC subunit-encoding genes in our transcriptomic analysis. A number of NF-Y trimers formed with MtNF-YA1 have been reported recently (Baudin et al., 2015). It is thus conceivable that the specificity of MtNF-YA1 functions in symbiotic and pathogenic plant microbe interactions is achieved by the formation of different NF-Y trimmers. Also, it is likely that MtNF-YA1 functions redundantly in the interaction with A. euteiches. This view is supported by the resistance triggered by miR169q overexpression in F83005.5 which is stronger than the one obtained via RNA interference. miR169 is encoded by a large gene family in plants (14 genes in Arabidopsis leading to four different mature miR169s) and targets specifically the product of NF-YA genes with a certain degree of specificity (Sorin et al., 2014). It is thus likely that other NF-YA subunits such as MtNF-YA2 for example that is expressed in roots at a much higher level than MtNF-YA1 are targeted by the miR169q in M. truncatula and play overlapping functions with MtNF-YA1 in the susceptibility to A. euteiches.

Soils harbor diverse microbes engaging in contrasting interactions with plant roots (Rey and Schornack, 2013). Legumes establish mutualistic symbiosis with rhizobacteria or arbuscular mycorrhizal fungi or activate immune processes to fend off pathogens. How selectivity is achieved by the host plant to build up appropriate immune and symbiotic responses stays elusive, especially since many similarities exist in the colonization strategies of beneficial and detrimental partners (Evangelisti et al., 2014). In M. truncatula, NF-YA1 knock out does not cause major developmental phenotypes for the plant but subtly modifies the associations with root inhabiting microbes. Knocking out MtNF-YA1 to enhance resistance to A. euteiches, a major threat for pea and alfafa (Gaulin et al., 2007) is an appealing approach to engineer durable resistance (Le Fevre et al., 2015) but will preclude nitrogen fixing symbiosis. However, genome editing (Feng et al., 2013) in non-legume crops for orthologs of MtNF-YA1 may become a fruitful biotech application of our findings.