Importance of the C12 Carbon Chain in the Biological Activity of Rhamnolipids Conferring Protection in Wheat against Zymoseptoria tritici

The hemibiotrophic fungus Zymoseptoria tritici, responsible for Septoria tritici blotch, is currently the most devastating foliar disease on wheat crops worldwide. Here, we explored, for the first time, the ability of rhamnolipids (RLs) to control this pathogen, using a total of 19 RLs, including a natural RL mixture produced by Pseudomonas aeruginosa and 18 bioinspired RLs synthesized using green chemistry, as well as two related compounds (lauric acid and dodecanol). These compounds were assessed for in vitro antifungal effect, in planta defence elicitation (peroxidase and catalase enzyme activities), and protection efficacy on the wheat-Z. tritici pathosystem. Interestingly, a structure-activity relationship analysis revealed that synthetic RLs with a 12 carbon fatty acid tail were the most effective for all examined biological activities. This highlights the importance of the C12 chain in the bioactivity of RLs, likely by acting on the plasma membranes of both wheat and Z. tritici cells. The efficacy of the most active compound Rh-Est-C12 was 20-fold lower in planta than in vitro; an optimization of the formulation is thus required to increase its effectiveness. No Z. tritici strain-dependent activity was scored for Rh-Est-C12 that exhibited similar antifungal activity levels towards strains differing in their resistance patterns to demethylation inhibitor fungicides, including multi-drug resistance strains. This study reports new insights into the use of bio-inspired RLs to control Z. tritici.


Introduction
Wheat is one of the most produced and consumed crops worldwide along with maize and rice. For the 2018-2019 period, the Food and Agriculture Organisation estimated that global wheat production reached 732.1 million tons [1]. This crop is cultivated in many different geoclimatic areas and ecosystems, and it is vastly used for human food as well as livestock feed. Hence, ensuring a safe and sufficient production of this cereal is a vital matter. Although wheat is susceptible to various pathogenic agents, Septoria tritici blotch (STB), caused by the hemibiotrophic fungus Zymoseptoria tritici, is considered as the most impacting foliar disease on wheat and is of major concern for wheat production worldwide. Currently, strategies to control this phytopathogen rely mainly on the application of synthetic environmentally-friendly RLs produced by green chemistry were found to be able to trigger ROS production in A. thaliana, showing a promising potential as plant resistance inducers [24]. Moreover, Robineau et al. [25] demonstrated that synthetic mono-RLs confer protection to tomato against B. cinerea by both direct antifungal activity and host defense mechanism activation. So far, biological activity of RLs has never been explored in the wheat-Z. tritici pathosystem. The objective of the present study was thus to assess the direct antifungal activity, the plant eliciting effect, as well as the protection efficacy of 19 RLs (either natural, produced by P. aeruginosa, or synthetic (bio-inspired) obtained by green chemistry), as well as lauric acid and dodecanol (C 12 -carbon chain molecules) on the wheat-Z. tritici model. A structure-activity relationship analysis of these compounds was performed in order to highlight the chemical radicals responsible for the RL activities. Moreover, the efficacy of the most active RL was compared between in vitro (direct antifungal activity) and in planta (disease symptom reduction) conditions. Finally, a possible strain-dependant effect of the RL activity was investigated by examining the in vitro antifungal activity of the most active RL toward 21 Z. tritici strains that differ in their resistance level to DMI fungicides.

Biological Activities of RLs Differ According to the Length of Their Carbon Chain
A total of 21 molecules, including 18 synthesized RLs and three related molecules (a mixture of natural RLs, lauric acid, and dodecanol) were examined for both their direct (antifungal) and indirect (plant defence elicitation) activities, as well as for their protection efficacy in the greenhouse, on the wheat-Z. tritici pathosystem. Regarding the in vitro antifungal activity, among all tested RLs, from C4 to C18-derived molecules, only Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12 drastically inhibited fungal growth (Table 2, Figure 1). Succinate-RLs did not significantly reduce fungal growth, except for Rh-Suc-C8, which showed an inhibitory effect only at the highest tested concentration of 1500 μM ( Table 2). Concerning the ester-RL family, Rh-Est-C10 had no antifungal activity, while Rh-Est-C12 was found to be the most active molecule among all tested RLs (Figure 1). On the other hand, lauric acid and dodecanol displayed a marked in vitro antifungal effect againt Z. tritici, with IC50 and MIC values close to those of the active RLs. Natural RLs did not show any direct antimicrobial effect.
Concerning the ability of the molecules to elicit plant defenses, activity of two plant enzymes involved in ROS scavenging, catalase (CAT) and peroxidase (POX), was investigated two days after treatment. Only three molecules significantly modified at least one enzymatic activity ( Table 2). Rh-Eth-C12 increased CAT activity by 1.86 fold compared to the control, whereas Rh-Suc-C12b enhanced POX activity by 2.26 fold compared to the control. Rh-Suc-C12, whose chemical structure is close to Rh-Suc-C12b,

Biological Activities of RLs Differ According to the Length of Their Carbon Chain
A total of 21 molecules, including 18 synthesized RLs and three related molecules (a mixture of natural RLs, lauric acid, and dodecanol) were examined for both their direct (antifungal) and indirect (plant defence elicitation) activities, as well as for their protection efficacy in the greenhouse, on the wheat-Z. tritici pathosystem. Regarding the in vitro antifungal activity, among all tested RLs, from C4 to C18-derived molecules, only Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12 drastically inhibited fungal growth (Table 2, Figure 1). Succinate-RLs did not significantly reduce fungal growth, except for Rh-Suc-C8, which showed an inhibitory effect only at the highest tested concentration of 1500 μM (Table 2). Concerning the ester-RL family, Rh-Est-C10 had no antifungal activity, while Rh-Est-C12 was found to be the most active molecule among all tested RLs (Figure 1). On the other hand, lauric acid and dodecanol displayed a marked in vitro antifungal effect againt Z. tritici, with IC50 and MIC values close to those of the active RLs. Natural RLs did not show any direct antimicrobial effect.
Concerning the ability of the molecules to elicit plant defenses, activity of two plant enzymes involved in ROS scavenging, catalase (CAT) and peroxidase (POX), was investigated two days after treatment. Only three molecules significantly modified at least one enzymatic activity (Table 2). Rh-Eth-C12 increased CAT activity by 1.86 fold compared to the control, whereas Rh-Suc-C12b enhanced POX activity by 2.26 fold compared to the control. Rh-Suc-C12, whose chemical structure is close to Rh-Suc-C12b,

Biological Activities of RLs Differ According to the Length of Their Carbon Chain
A total of 21 molecules, including 18 synthesized RLs and three related molecules (a mixture of natural RLs, lauric acid, and dodecanol) were examined for both their direct (antifungal) and indirect (plant defence elicitation) activities, as well as for their protection efficacy in the greenhouse, on the wheat-Z. tritici pathosystem. Regarding the in vitro antifungal activity, among all tested RLs, from C4 to C18-derived molecules, only Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12 drastically inhibited fungal growth (Table 2, Figure 1). Succinate-RLs did not significantly reduce fungal growth, except for Rh-Suc-C8, which showed an inhibitory effect only at the highest tested concentration of 1500 μM (Table 2). Concerning the ester-RL family, Rh-Est-C10 had no antifungal activity, while Rh-Est-C12 was found to be the most active molecule among all tested RLs (Figure 1). On the other hand, lauric acid and dodecanol displayed a marked in vitro antifungal effect againt Z. tritici, with IC50 and MIC values close to those of the active RLs. Natural RLs did not show any direct antimicrobial effect.
Concerning the ability of the molecules to elicit plant defenses, activity of two plant enzymes involved in ROS scavenging, catalase (CAT) and peroxidase (POX), was investigated two days after treatment. Only three molecules significantly modified at least one enzymatic activity (Table 2). Rh-Eth-C12 increased CAT activity by 1.86 fold compared to the control, whereas Rh-Suc-C12b enhanced POX activity by 2.26 fold compared to the control. Rh-Suc-C12, whose chemical structure is close to Rh-Suc-C12b,

Biological Activities of RLs Differ According to the Length of Their Carbon Chain
A total of 21 molecules, including 18 synthesized RLs and three related molecules (a mixture of natural RLs, lauric acid, and dodecanol) were examined for both their direct (antifungal) and indirect (plant defence elicitation) activities, as well as for their protection efficacy in the greenhouse, on the wheat-Z. tritici pathosystem. Regarding the in vitro antifungal activity, among all tested RLs, from C4 to C18-derived molecules, only Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12 drastically inhibited fungal growth (Table 2, Figure 1). Succinate-RLs did not significantly reduce fungal growth, except for Rh-Suc-C8, which showed an inhibitory effect only at the highest tested concentration of 1500 µM (Table 2). Concerning the ester-RL family, Rh-Est-C10 had no antifungal activity, while Rh-Est-C12 was found to be the most active molecule among all tested RLs ( Figure 1). On the other hand, lauric acid and dodecanol displayed a marked in vitro antifungal effect againt Z. tritici, with IC 50 and MIC values close to those of the active RLs. Natural RLs did not show any direct antimicrobial effect.
Concerning the ability of the molecules to elicit plant defenses, activity of two plant enzymes involved in ROS scavenging, catalase (CAT) and peroxidase (POX), was investigated two days after treatment. Only three molecules significantly modified at least one enzymatic activity (Table 2). Rh-Eth-C12 increased CAT activity by 1.86 fold compared to the control, whereas Rh-Suc-C12b enhanced POX activity by 2.26 fold compared to the control. Rh-Suc-C12, whose chemical structure is close to Rh-Suc-C12b, did not exhibit any induction of the targeted enzymes in our conditions. Finally, Rh-Est-C12 induced both CAT activity and POX activities by 1.71 and 2.27 folds, respectively, when compared to the control. Lauric acid and dodecanol did not cause any significant change in the investigated enzyme activities.

Correlations between In Vitro Antifungal Activity, In planta Defence Elicitation and Protection Efficacy
A correlative analysis was performed to determine whether in vitro antifungal and elicitation activities were relevant to explain disease symptom reduction caused by the treatments. The statistical analyses revealed significant positive correlations between both IC50 and MIC and the disease severity reduction (r = 0.46) ( Table 3). However, no significant correlation was found between CAT activity and protection efficacy, while POX activity and disease symptoms were significantly and negatively correlated (r = −0.51). Table 3. Correlations between protection efficacy and both direct (IC50 and MIC) and indirect (plant catalase and peroxidase activities) effects of the tested RLs, as well as dodecanol and lauric acid, obtained on the wheat-Zymoseptoria tritici pathosystem.

Comparison of In Vitro Versus in Planta Antifungal Effects of Rh-Est-C12
According to the structure-activity relationship study above, Rh-Est-C12 appeared to be the most active RL for protection efficacy, in vitro antifungal effect, and elicitation activity. Hence, we focused on this molecule in order to compare its direct activity in vitro and in planta. In planta, the protective effect of foliar treatment with Rh-Est-C12 on wheat against Z. tritici was assessed at different concentrations. There was no significant protection with applications of Rh-Est-C12 at 50, 100, 200, and 400 μM of the RL. However, Rh-Est-C12 treatments with 800, 1600, and 3200 μM significantly reduced disease  Table 2. Biological activities, including in vitro direct antifungal activity, in planta defense activation and in planta protection efficacy of the eighteen synthetized RLs, as well as lauric acid, dodecanol and natural RLs, at 1.5 mM, on the wheat-Zymoseptoria tritici pathosystem.

In Vitro Antifungal Activity
In None of the tested molecule displayed in planta phytotoxicity at the tested concentration (1.5 mM). Five of the tested RLs caused a significant reduction in disease symptoms (Table 2). Among them, the four molecules Rh-Eth-C12, Rh-Suc-C12b, Rh-Eth-C10, and Natural-Rh conferred only a slight protection effect, with a decrease in disease severity ranging from 15.5 to 17.4%. On the other hand, disease severity reduction obtained with Rh-Est-C12 treatment was significantly different from all other treatments and exhibited the highest activity, with 35.6% of disease symptom reduction (approximatively 2 fold higher than the protection efficacies conferred by the other active RLs). On the contrary to RLs with C12-carbon chain, lauric acid and dodecanol did not display any significant protection activity.

Correlations between In Vitro Antifungal Activity, in Planta Defence Elicitation and Protection Efficacy
A correlative analysis was performed to determine whether in vitro antifungal and elicitation activities were relevant to explain disease symptom reduction caused by the treatments. The statistical analyses revealed significant positive correlations between both IC 50 and MIC and the disease severity reduction (r = 0.46) ( Table 3). However, no significant correlation was found between CAT activity and protection efficacy, while POX activity and disease symptoms were significantly and negatively correlated (r = −0.51).

Comparison of In Vitro Versus in Planta Antifungal Effects of Rh-Est-C12
According to the structure-activity relationship study above, Rh-Est-C12 appeared to be the most active RL for protection efficacy, in vitro antifungal effect, and elicitation activity. Hence, we focused on this molecule in order to compare its direct activity in vitro and in planta. In planta, the protective effect of foliar treatment with Rh-Est-C12 on wheat against Z. tritici was assessed at different concentrations. There was no significant protection with applications of Rh-Est-C12 at 50, 100, 200, and 400 µM of the RL. However, Rh-Est-C12 treatments with 800, 1600, and 3200 µM significantly reduced disease symptoms on wheat leaves, with protection rates of 19.9%, 53.3%, and 78.9% respectively (Figure 2). No phytotoxicity was observed at any of the tested concentrations. The in planta IC 50 value of this molecule (the concentration of Rh-Est-C12 reducing by half the disease symptom level compared to the non-treated control) was calculated and was determined as 1510 µM. Rh-Est-C12 exhibited an in planta IC 50 value approximatively 20-fold higher than the IC 50 value obtained in vitro for the molecule, revealing a major difference in the activity of the molecule between the two conditions.
Additionally, the effect of Rh-Est-C12 on in planta spore germination and fungal hyphal growth was assessed at three different concentrations (50, 400, and 3200 µM) (Figure 3). Whereas no significant activity was observed after application of Rh-Est-C12 at 50 and 400 µM, a strong antifungal effect was highlighted for this compound at 3200 µM. The proportion of non-germinated spores in the treated plants was 2-fold higher than in the control. Moreover, a reduction by 4-fold of the class 4 spore proportion (geminated spores with a strongly developed germ tube) was also scored in the plants treated with 3200 µM Rh-Est-C12 compared to the control (Figure 3). Again, the in planta direct activity of Rh-Est-C12 on hyphal growth appeared weaker than its direct antifungal effect obtained in vitro on fungal colony development.
symptoms on wheat leaves, with protection rates of 19.9%, 53.3%, and 78.9% respectively (Figure 2). No phytotoxicity was observed at any of the tested concentrations. The in planta IC50 value of this molecule (the concentration of Rh-Est-C12 reducing by half the disease symptom level compared to the non-treated control) was calculated and was determined as 1510 μM. Rh-Est-C12 exhibited an in planta IC50 value approximatively 20-fold higher than the IC50 value obtained in vitro for the molecule, revealing a major difference in the activity of the molecule between the two conditions. Additionally, the effect of Rh-Est-C12 on in planta spore germination and fungal hyphal growth was assessed at three different concentrations (50, 400, and 3200 μM) (Figure 3). Whereas no significant activity was observed after application of Rh-Est-C12 at 50 and 400 μM, a strong antifungal effect was highlighted for this compound at 3200 μM. The proportion of non-germinated spores in the treated plants was 2-fold higher than in the control. Moreover, a reduction by 4-fold of the class 4 spore proportion (geminated spores with a strongly developed germ tube) was also scored in the plants treated with 3200 μM Rh-Est-C12 compared to the control ( Figure 3). Again, the in planta direct activity of Rh-Est-C12 on hyphal growth appeared weaker than its direct antifungal effect obtained in vitro on fungal colony development.

Direct In Vitro Antifungal Activity of Rh-Est-C12 Does not Vary Depending on Z. tritici Strains
To examine any possible strain effect of Rh-Est-C12 direct in vitro antifungal activity, 21 different Z. tritici strains showing different resistance levels towards DMI fungicides, were investigated for their resistance level to this RL. Overall, no major difference was observed between all tested strains regarding the IC50 values of Rh-Est-C12 ( Figure 4A,   Figure 3. In planta spore germination and hyphal growth of Zymoseptoria tritici on wheat leaves treated or not with Rh-Est-C12. Wheat cultivar Alixan and Z. tritici strain T02596 were used. Percentage of four different classes of Calcofluor-strained spores was assessed three days after inoculation on wheat third leaves. The right panel shows a representative picture of each spore class (scale bar = 25 µm). Class 1: non germinated spore; Class 2: geminated spore with small germ tube; Class 3: geminated spore with developed germ tube; Class 4: geminated spore with a strongly developed germ tube. For each condition, 100 different spores from six different leaves were assessed. Within each class, different letters indicate significant differences between treatments according to ANOVA followed by the Tukey's test at p ≤ 0.05.

Direct In Vitro Antifungal Activity of Rh-Est-C12 Does not Vary Depending on Z. tritici Strains
To examine any possible strain effect of Rh-Est-C12 direct in vitro antifungal activity, 21 different Z. tritici strains showing different resistance levels towards DMI fungicides, were investigated for their resistance level to this RL. Overall, no major difference was observed between all tested strains regarding the IC 50 values of Rh-Est-C12 ( Figure 4A, Tables S1 and S2). Resistance factors (ratio between the IC 50 value of a given strain and the IC 50 value of the sensitive reference strain IPO323) ranged from 0.7 to 1.2 for Rh-Est-C12, indicating no strain effect on the direct activity of the molecule. By contrast, a strong strain effect was highlighted for the five tested DMI fungicides, with resistance factors ranging from 3.8 to 1071.4 for tebuconazole, from 1.0 to 150.8 for metconazole, from 28.6 to 501.8 for epoxiconazole, 37.4 to 74753.2 for prothioconazole, and from 5.1 to 266.2 for prochloraz ( Figure 4A, Tables S1 and S2). Rh-Est-C12 displayed similar antifungal activity levels on both (i) non-MDR and non Cyp51-overexpressing strains, and (ii) the highly resistant strains with MDR character or overexpressing the Cyp51 gene ( Figure 4B, Tables S1 and S2).  or distinguished according to their DMI resistance profiles (seven MDR strains, seven strains overexpressing the Cyp51 gene, and seven strains without these resistance mechanisms) (b), against five DMI fungicides as well as the RL molecule Rh-Est-C12. Resistance factors correspond to the ratio between the IC50 value of a given strain and the IC50 value of the sensitive reference strain IPO323.

Discussion
We explored here, for the first time, the potential of RLs to protect wheat against Z. tritici, a major pathogen on wheat crops, using a panel of 21 molecules, including 18 bioinspired RLs synthesized using green chemistry and three related molecules (a mixture of natural RLs from P. aeruginosa, lauric acid, and dodecanol). Natural bacterial RLs have already been reported to show antimicrobial activity against a wide range of plant pathogenic fungi [13]. For instance, Varnier et al. [17] reported a significant antigerminative effect of 0.1 mg·mL -1 of bacterial RL against the ascomycete polyphagous fungus B. cinerea. Moreover, natural RLs were found particularly effective against certain oomycetes, such as Phytophthora sp., by lysing zoospores, hypothetically by intercalating into oomycete cell membranes, leading to their disorganization [14]. However, in our conditions, and even with concentrations up to 0.87 mg·mL −1 (i.e., 1500 μM) no activity was observed for the tested mixture of natural RLs. Among all tested compounds, only a minority exhibited significant in vitro antifungal activity. The structure-activity relationship analysis revealed that bioinspired RLs with 4, 6, 8, 14, 16, and 18 carbon fatty acid tails did not display any significant direct effect. Similarly, succinate-derived molecules were also unable to inhibit fungal in vitro growth, except Rh-Suc-C8 which exhibited a very slight effect. The most effective RLs were ether and ester-derived compounds, namely Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12. Notably, Robineau et al. [25] also described the antifungal activity of these three compounds against B. cinerea.

Discussion
We explored here, for the first time, the potential of RLs to protect wheat against Z. tritici, a major pathogen on wheat crops, using a panel of 21 molecules, including 18 bioinspired RLs synthesized using green chemistry and three related molecules (a mixture of natural RLs from P. aeruginosa, lauric acid, and dodecanol). Natural bacterial RLs have already been reported to show antimicrobial activity against a wide range of plant pathogenic fungi [13]. For instance, Varnier et al. [17] reported a significant antigerminative effect of 0.1 mg·mL −1 of bacterial RL against the ascomycete polyphagous fungus B. cinerea. Moreover, natural RLs were found particularly effective against certain oomycetes, such as Phytophthora sp., by lysing zoospores, hypothetically by intercalating into oomycete cell membranes, leading to their disorganization [14]. However, in our conditions, and even with concentrations up to 0.87 mg·mL −1 (i.e., 1500 µM) no activity was observed for the tested mixture of natural RLs. Among all tested compounds, only a minority exhibited significant in vitro antifungal activity. The structure-activity relationship analysis revealed that bioinspired RLs with 4, 6, 8, 14, 16, and 18 carbon fatty acid tails did not display any significant direct effect. Similarly, succinate-derived molecules were also unable to inhibit fungal in vitro growth, except Rh-Suc-C8 which exhibited a very slight effect. The most effective RLs were ether and ester-derived compounds, namely Rh-Eth-C10, Rh-Eth-C12, and Rh-Est-C12. Notably, Robineau et al. [25] also described the antifungal activity of these three compounds against B. cinerea. However, they observed a more pronounced effect with Rh-Eth-C10. In the present work, Rh-Eth-C10 was not as effective as Rh-Est-C12 that displayed the lowest IC 50 among all the tested active compounds (158 and 75 µM, corresponding to 0.048 and 0.026 mg·mL −1 , for Rh-Eth-C10 and Rh-Est-C12, respectively) (Table 1). Moreover, Rh-Est-C10 did not exhibit any inhibition of fungal growth. Hence, we conclude that 12 carbon is the optimum fatty acid tail length for both ether and ester-RLs regarding their direct activity against Z. tritici. Thereafter, we assessed the antifungal potential of chemically simple molecules with 12 carbon chain, lauric acid and dodecanol. Interestingly, these two molecules exhibited a significant direct effect, but their IC 50 are higher than those of Rh-Est-C12 and Rh-Eth-C12 (Table 2). These results confirm the importance of the 12-carbon chain in the direct activity of RLs towards Z. tritici.
We also investigated the elicitor potential of RLs on non-infected wheat using two key enzyme biomarkers of plant defence. Only three molecules significantly enhanced the activity of at least one out of the two targeted plant defense-related enzymes CAT and POX. Rh-Eth-C12 was able to stimulate CAT activity, Rh-Suc-C12b induced POX activity, and Rh-Est-C12 increased both (Table 2). These enzymes are involved in many plant defense responses including ROS scavenging, plant signaling, and hypersensitive response [26,27]. These results suggest an increase of ROS production in wheat plants treated with these three compounds. Since large amounts of ROS in plant tissues can be harmful to the plant, wheat plants likely responded by increasing the activity of ROS scavenging enzymes. Interestingly, natural and synthetic RLs were reported to confer resistance and trigger oxidative burst in other plants such as A. thaliana and grapevine [17,24]. Robineau et al. [25] also described a significant increase of ROS production in A. thaliana leaf disks treated with Rh-Eth-C12 and Rh-Est-C12. Accumulation of H 2 O 2 in wheat leaves during infection with Z. tritici was reported to increase plant resistance against the pathogen [28]. While the role of CAT in wheat defense against Z. tritici is still unclear, Shetty and colleagues [29] reported an involvement of POX in wheat innate immunity. The authors observed a significant induction of this enzyme in resistant cultivars when infected with Z. tritici, compared to susceptible cultivars. This may be explained by the specific roles of POX, such as the reinforcement of cell walls and deposition of polyphenolics [30]. This link between POX and innate resistance of wheat to Z. tritici is supported by the positive correlation scored between the induction of POX enzymatic activity and plant protection, i.e., negative correlation between POX activity and disease symptoms ( Table 3).
As for the in vitro direct antifungal activity, interestingly, our results revealed that RLs with C12 fatty acid chain were the most effective regarding the induction of the targeted wheat defense biomarkers (CAT and POX). This dual activity on both the host (wheat) and the pathogen (Z. tritici) is more likely due to RL interaction with biological plasma membranes of both organisms. Indeed, Monnier and colleagues [31] demonstrated, using in silico models, that RLs can act on both plant and fungal membranes by insertion within the lipid bilayers, near the lipid phosphate group of the phospholipid bilayers, nearby phospholipid glycerol backbones. Even though RLs fit into both membranes, RLs seem to differently affect their fluidity. Indeed, the authors suggested that, in the plant membrane, RLs did not significantly impact lipid dynamics inside the lipid bilayer, whereas, in the fungal membrane, a clear fluidity increase was observed, due probably to the presence of ergosterol. These subtle changes in lipid dynamics could explain why the effect of RLs is less drastic on the plant (only triggering defense reactions) than on the fungus that leads to a strong membrane destabilization causing cell death [31]. The importance of the RL carbon chain in plant defense activation has already been reported in previous studies. For instance, Luzuriaga-Loaiza et al. [24] highlighted different results with synthetic RL bolaforms (SRBs) in A. thaliana, where SRBs with 14-carbon fatty acid tails were found to be the most effective to induce ROS production in the plant petioles. Regarding other surfactant agents, it has been shown that for surfactin, a lipopeptide produced by Bacillus subtilis, the molecules with chain length of 14 or 15 carbons were found more active than surfactin with 12 carbons to enhance production of H 2 O 2 in tobacco cells [32]. On the other hand, we found that lauric acid and dodecanol were not able to elicit POX-and CAT-based defense reactions, suggesting a major role of the sugar head in plant defense induction. Although further investigations are needed to decipher the modes of action of RLs on wheat defense induction, such a result suggests the major importance of both rhamnose head and fatty acid tail in the biological activities of RLs on this plant model. Moreover, the precise structure of the RL carbon chain is also crucial, since we observed a significant enhancement of POX activity when using Rh-Suc-C12b compared to its isoform Rh-Suc-C12, which showed no elicitation activity in the treated wheat plants.
Our results revealed that Rh-Eth-C10, Rh-Eth-C12, Rh-Est-C12, Rh-Suc-C12b, and natural RLs are able to reduce Z. tritici disease severity on wheat plants. Overall, four categories of protective RLs were observed: (i) protective RLs exhibiting only antifungal activity (Rh-Eth-C10); (ii) protective RLs displaying only elicitation effect (Rh-Suc-C12b); (iii) protective RLs with both activities (Rh-Eth-C12 and Rh-Est-C12), and (iv) protective RLs displaying neither direct nor indirect effects (Natural-Rh), implying that this latter compound could induce plant defenses through mechanisms not involving our two targeted enzymatic biomarkers. Statistical analyses revealed a positive correlation between POX activity and protection efficacy, as well as between in vitro direct activity and protection efficacy. Surprisingly, even though lauric acid and dodecanol were able to inhibit fungal growth in vitro with IC 50 values closes to those of protective RLs, they were unable to protect wheat against the disease, highlighting again the role of rhamnose head in RL protective activity. As discussed above, our structure-activity relationship study reveals the predominant importance of 12 carbon chain in RL biological activities in the wheat-Z. tritici model. Likewise, Robineau et al. [25] also described the important role of 12 carbon fatty acid tails in the bioactivity of RLs on the A. thaliana-B. cinerea pathosytem.
The mono-RL Rh-Est-C12 was found to be the most effective RL to protect wheat against Z. tritici, displaying a reduction of disease symptoms up to 78.9% when applied at 1.1 mg·mL −1 (i.e., 3200 µM). This product is able to act dually, by inhibiting fungal growth and eliciting plant defenses by enhancing ROS scavenging enzyme activities. Nevertheless, a significant gap between the in vitro and in planta efficacies of this molecule was highlighted. A ratio of approximately 20-fold separates the in vitro IC 50 value of Rh-Est-C12 and the needed concentration to reduce disease symptoms by 50% in planta. This difference of activity may be explained by several factors, including the adherence level of Rh-Est-C12 on wheat leaves during spraying, biodegradability as well as photodecomposition, and penetration properties of the molecule in wheat leaves. Further studies on formulation may increase and optimize the in planta efficacy of Rh-Est-C12. Another explanation of the gap between in vitro and in planta activities of Rh-Est-C12 is the difference between the physiological state of the fungus in vitro and in planta. Indeed, the fungus develops in planta hyphae which can be considered as more robust (since it is in its wild environment, i.e., within its host) than the yeast-like morphotype formed by Z. tritici in vitro. Another major difference between in vitro and in planta conditions is that Z. tritici develops in planta several specialized structures such as pseudo-appressoria and pycnidia that it cannot form in vitro, because of the absence of the host in this latter condition.
Considering the high diversity of Z. tritici populations that can be found in a single geographic area [33], we assessed the variability of the direct antifungal activity of Rh-Est-C12 towards 21 different strains of the fungus differing in their resistance level to DMI fungicides, the most used chemical class in the field to control this pathogen. Although the in vitro IC 50 value of Rh-Est-C12 was, as expected, much higher than those of conventional fungicides on the reference strain IPO323 (54.3 µg·mL −1 , i.e., 157 µM, for the RL vs. 0.001 to 0.07 µg·mL −1 depending on the considered fungicide, Table S1), it is notherworthy that no strain effect was observed for Rh-Est-C12. By contrast, strong resistance factors were scored for all tested DMI molecules, thus confirming previous studies reporting high rates of DMI resistance in Z. tritici populations [34]. Hence, we could assume that Rh-Est-C12 display a mode of action on Z. tritici different from that of DMI fungicides, which inhibit the sterol 14 α-demethylase, CYP51, preventing the fungus to produce ergosterol [35]. Here, we suggest that Rh-Est-C12 interacts directly with Z. tritici cell membranes, resulting in an inhibition of the fungal growth, as previously suggested for other fungi [14,18].
The environmentally-friendly properties of RLs, the duality of Rh-Est-C12 biological activities (direct and indirect activities), an absence of strain specific resistance within the examined Z. tritici collection, together with the low-cost production and industrial feasibility of bioinspired synthetic RLs compared to bacterial production and purification, makes Rh-Est-C12 a promising candidate to control Z. tritici in the field. Nevertheless, additional investigations on the formulation aiming at reducing the concentration of the compound in planta are required.
The conversions and the purities of the synthetized products were determined by NMR spectroscopy. NMR spectra were recorded on a DRX300 spectrometer (Bruker, Wissembourg, France) operating at 300 MHz for 1 H nuclei and 75 MHz for 13 C nuclei. CDCl 3 (99.50% isotopic purity) were purchased from Euriso-Top (Saint Aubin, France). The progress of the reactions was checked by thin layer chromatography (TLC) on silica gel 60 glass plates (Merck, Fontenay-sous-Bois, France). Detection was carried out by spraying the chromatograms with 10% ethanolic sulfuric acid and heating them to 100 • C. Flash column chromatography was performed with silica gel (40-100 µm, Merck): all chemicals were of reagent grade and used without further purification.
The general procedures for rhamnose ether, ester and monorhamnosyl (alkenyl)succin -ate molecule syntesesis have been reported in Robineau et al. [25], while the general procedure for monorhamnosyl (alkenyl)alkylsuccinate derivative synthesis will be described as follow. First, alkenylsuccinic anhydride (12.2 mmol, 1 eq.) was added dropwise to a stirred solution of L -rhamnose (2.0 g, 12.2 mmol, 1 eq.) and DMAP (1.8 g, 14.6 mmol, 1.2 eq.) in DMF (20 mL). After 16 h at 60 • C, DMAP (1.5 g, 12.2 mmol, 1 eq.) and alkyl bromide (14.6 mmol, 1.2 eq.) was added slowly to the reaction mixture and stirred overnight. The reaction was cooled and H 2 SO 4 10% was added to reached pH 5. Ethyl acetate was added and the organic phase was washed with a saturated NaCl solution and HCl 1 M. The organic layer was dried with MgSO 4 and concentrated under reduced pressure. The rhamnose alkylsuccinate was isolated after purification by column chromatography (EtOAc/Petroleum spirit 5:5).

RL Chemical Analysis
The chemical analyses of the previously reported RLs Rh-Eth-C4, Rh-Eth-C6, Rh-Eth-C8, Rh-Eth-C10, Rh-Eth-C12, Rh-Eth-C14, Rh-Eth-C16, Rh-Est-C12, Rh-Suc-C12 were already performed [25]. For the nine new synthesized RLs in the present study, their chemical analyses are described below. Confirmatory analyses included elemental analysis (vario MICRO cube CHNS/O, Elementar, Lyon, France); and nuclear magnetic resonance spectroscopy (NMR, Bruker 400 MHz spectrometer).  third-leaves are fully expanded, 30 mL of a solution of 1.5 mM of each RL, as well as lauric acid and dodecanol, were hand-sprayed on the plants of each pot. These molecules were first dissolved in dimethyl sulfoxide (DMSO), with final concentration of DMSO at 0.2%, before being supplemented with 0.05% polyoxyethylene-sorbitan monolaurate (Tween 20, Sigma-Aldrich, Saint-Quentin-Fallavier, France) as a wetting agent. Control plants were treated with a solution of 0.2% DMSO, supplemented with 0.05% Tween 20. Due to solubilization difficulties at 1.5 mM, treatments with Rh-Eth-C18 and Rh-Suc-C12-C12 were not performed. Two days after treatment, 30 mL of fungal spore suspension (10 6 spores·mL −1 ), also supplemented with 0.05% of Tween 20, were hand-sprayed on the plants of each pot. Zymoseptoria tritici spores were harvested after being cultivated on potato dextrose agar medium (PDA) plates for one week in dark conditions, according to Siah et al. [37]. Immediately after inoculation, each pot was covered with a clear polyethylene bag for 3 days in order to ensure a high humidity level allowing a good plant infection. Three weeks after inoculation, the disease severity was measured on the third leaf by assessing the area covered by lesions (chlorosis or necrosis) bearing or not pycnidia. This experiment was repeated twice. In the case of Rh-Est-C12, the most efficient RL in our conditions, the protection efficacy of this compound was assessed at different concentrations, ranging from 50 µM to 3200 µM (50, 100, 200, 400, 800, 1600, and 3200 µM). This experiment underwent the same conditions as described above, except for DMSO that was used at 0.1% final concentration in the treatment solutions.

Plant Defence-Related Enzyme Assays
Evaluation of the indirect effect of RLs, as well as lauric acid and dodecanol, was performed by assessing the activity of two antioxidant key-enzymes involved in plant defense mechanisms: catalase (CAT; EC: 1.11.1.6) and peroxidase (POX; EC: 1.11.1.7). Wheat plants grown under greenhouse conditions treated or not with 1.5 mM of RLs, as described above, were used to examine the elicitation activity of the compounds. Plant third-leaves were harvested two days after treatment and 100 mg were immediately frozen in liquid nitrogen and stored at −80 • C for further analyses. Three leaves, sampled randomly from three different pots (one leaf per pot), were used as repetitions for each condition. Leaf segments were later ground with a mortar and pestle in liquid nitrogen. The powder obtained was resuspended in 1 mL of 50 mM ice-cold potassium phosphate buffer, pH 7.0, and centrifuged at 12,000× g for 10 min at 4 • C. The supernatant was used to evaluate the activity of CAT or POX. POX activity was assessed based on the procedure of Mitchell et al. [38] and Shindler et al. [39]. The reactive medium consisted of 0.5 mM 2,2 -azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) and 0.25 mM H 2 O 2 in 25 mM sodium acetate buffer, pH 4.4. This assay was performed in flat-bottomed polystyrene 96-well microplates (Corning Costar ® , Corning, NY, USA). To initiate the reaction, 10 µL of supernatant was added to 190 µL of the reactive mix in plate wells. The increased absorbance at 412 nm, produced by the formation of the radical cation, was monitored kinetically for 5 min at 25 • C. POX activity was expressed as change in absorbance at 412 nm·min −1 ·g −1 protein. CAT activity was determined according to Beers and Sizer [40] and García-Limones et al. [41] with slight modifications. To initiate the reaction, 10 µL of wheat leaf supernatant diluted in sterile water 1:4 (v/v) were added to the wells containing 190 µL of reactive medium consisting of 50 mM H 2 O 2 in 50 mM potassium phosphate buffer, pH 7.0. 96-well microplates (Greiner UV-Star ® , Greiner Bio-One GmbH, Kremsmünster, Austria), allowing measurements up to 200 nm, were used. The decreased absorbance at 240 nm resulting from the degradation of H 2 O 2 was measured kinetically during 5 min at 25 • C. CAT activity was expressed as change in absorbance at 240 nm·min −1 ·g −1 protein.
Protein content of wheat leaf supernatants was assessed using the method of Bradford [42] with bovin serum albumin (BSA) as a standard.

In Planta Antifungal Effect of Rh-Est-C12
Wheat plants grown under greenhouse conditions, sprayed or not with different concentrations of Rh-Est-C12 (0, 50, 400, and 3200 µM) and inoculated with spore suspension of the T02596 Z. tritici strain, were used to assess in planta the activity of this molecule on spore germination and hyphal growth of the phytopathogen. The three concentrations 50, 400, and 3200 µM were chosen because they correspond to the lowest, median, and highest concentrations tested in planta for this RL, respectively. The chitin staining dye Fluorescent Brightener 28 (Calcofluor, Sigma-Aldrich, USA) was used to stain the fungus on the leaf surfaces, according to Siah et al. [37]. Three days after inoculation, wheat third-leaf segments (2 cm) were harvested and immersed for five minutes in a solution of 0.1% (w/v) Calcofluor and 0.1 M Tris-HCl buffer, pH 8.5. Leaf segments were then washed twice in sterile osmosed water for two minutes. After being dried in dark conditions at room temperature, they were placed on a glass slide, covered with a cover slip and observed microscopically (Eclipse 80i, Nikon, Champigny-sur-Marne, France) under UV-light. For each leaf segment, the percentage of different spore classes was calculated on 100 observed spores (class 1: non germinated spore; class 2: geminated spore with a small germ tube; class 3: geminated spore with a developed germ tube; class 4: geminated spore with a strongly developed germ tube). Six third-leaf segments from three different pots (two segments per pot) were used as replicates for each condition. Pictures were taken with a digital camera (DXM1200C, Nikon, Champigny-sur-Marne, France) using image capture software (NIS-Elements BR, Nikon, Champigny-sur-Marne, France).

In Vitro Direct Antifungal Effect of Rhamnolipids
In vitro antifungal activity of all tested RLs, as well as dodecanol and lauric acid, was assessed by measuring the growth of Z. tritici colonies on solid medium in sterile 12-well plates (Cellstar standard ® , Greiner Bio-One GmbH, Kremsmünster, Austria). Plate wells were filled with 3 mL of PDA, autoclaved beforehand, mixed with our molecules of interest at approximatively 40 • C. The compounds were first dissolved in DMSO (0.1% final concentration in the wells). After PDA cooling down, 5 µL of the Z. tritici strain T02596 spore suspension (5 × 10 5 spores·mL −1 ) were added to each well, according to Siah et al. [37]. Ten different concentrations (39.5, 59.3, 88.9, 133.3, 200, 300, 450, 666.7, 1000, and 1500 mM) were used for each molecule. Moreover, control mediums supplemented with 0.1% of DMSO, with or without fungal inoculation, were used. After inoculation, plates were incubated for 10 days in growth chamber in dark conditions at 20 ± 1 • C, after which two perpendicular diameters of fungal colonies were measured visually. This experiment was repeated three times and three wells were used as replicates for each condition.

Fungal Strain Effect of Rh-Est-C12 Antifungal Activity
To assess the fungal strain effect of the most active RL Rh-Est-C12 regarding its antifungal direct activity, a total of 21 single-conidial Z. tritici strains (Zt.1 to Zt.21) isolated in 2016 from the Arras location in northern France, and the reference strain IPO323 isolated in The Netherlands in 1981 [43], were assessed for their variability in sensitivity to Rh-Est-C12. These Z. tritici strains were chosen because they displayed, in preliminary assays, different resistance levels towards five fungicide molecules belonging to DMI class (tebuconazole, metconazole, epoxiconazole, prothioconazole, prochloraz). The sensitivity of the strains towards these fungicides was evaluated in 96-well microplates, as described in Siah et al. [37], at the concentrations 75, 25, 8.33, 2.78, 0.93, 0.31, 0.10, and 0.03 mg·L −1 . The antifungal bioassay for Rh-Est-C12 against the 22 Z. tritici strains was performed at the concentrations 450, 300, 200, 133.3, 88.9, 59.3, and 39.5 µM.

Statistical Analyses
Data obtained for in planta protection efficacy, enzyme activity, and in planta spore germination assays were subjected to analysis of variance (ANOVA) followed by Tukey's test at p ≤ 0.05, using the XLSTAT software (Addinsoft, Paris, France). For in vitro antifun-gal activity assays, the minimal inhibitory concentration (MIC), defined as the minimal tested concentration totally inhibiting fungal growth, was scored for each molecule. In addition, half-maximal inhibitory concentration (IC 50 ) was calculated using GraphPad Prism software version 8 (GraphPad Software Inc., San Diego, CA, USA) for each molecule and tested fungal strain. In the strain effect assay, resistance factors were calculated by dividing the IC 50 of each strain by the IC 50 of the reference strain IPO323 for each tested compound. Correlations between protection efficacies, in vitro direct activity (IC 50 and MIC), and in planta indirect effect (CAT and POX activities) within all the tested molecules, was carried out with principal component analysis (PCA) based on the Pearson correlation test, using XLSTAT software (Addinsoft).