Introduction

The durable nature of partial or Quantitative Disease Resistance (QDR) in plants makes it a promising alternative to reduce the use of pesticides in agriculture. This is particularly true concerning pathogens with high evolutionary potential and necrotrophic pathogens1. Several genes control QDR and are associated with quantitative resistance loci (QRL) that each variably contribute to the phenotype2. Meanwhile, in qualitative disease resistance, R-genes are associated with total resistance to a disease. Their mechanism is generally based on the early detection of phytopathogen effectors. Specifically, R-genes facilitate a hypersensitivity response that causes the neighboring cells on the pathogen infection site to undergo programmed cell death3,4. The same response to necrotrophic pathogens would increase the plant’s susceptibility to the disease, as necrotrophic pathogens exploit cell death for their benefit. Hence, QDR appears to be a better means to mitigate the diseases caused by necrotrophic and hemi-biotrophic pathogens. However, in general, the levels of QDR obtained are insufficient to eliminate further phytosanitary measures such as the use of pesticides. Furthermore, breeding by exploiting QRLs is challenging as some of them can only be detected under certain environmental conditions or in specific genetic backgrounds2. As a result, observing QRL phenotypes and unveiling their underlying mechanisms have proven to be difficult.

No comprehensive model of QDR has been realized so far. Moreover, very different mechanisms of QDR have been uncovered, suggesting that such a unifying model does not exist1,5. Alternatively, it can be envisioned that several mechanisms concur to yield a general QDR level. These mechanisms include (i) morphological and developmental variations, (ii) basal defense involvement, (iii) chemical warfare, (iv) defense signal transduction pathways, and (v) QRLs as a weak version of R-genes1. Recently, other mechanisms underlying QDR have been described, including vesicle trafficking, molecular chaperoning, and detoxification5.

Phytotoxins are important determinants of plant diseases. Depending on their specificities, they can be classified as either non-host-specific toxins (NHSTs) or host-specific toxins (HSTs)6. Both NHSTs and HSTs fall under either of the two categories based on their chemical nature: ribosome synthesis-dependent peptides or secondary metabolites (SMs), which house most of the phytotoxins7. Several mechanisms of how phytotoxins are deployed have been described, including hindering lipid metabolism and disrupting plant membrane function. For instance, AAL toxins produced by the fungus Alternaria alternata are analogs of the plant’s ceramide synthase substrate. When exposed to this toxin, the plant cell undergoes rapid-fire production of sphingolipid precursors, leading to the loss of integrity of the plant’s plasma membrane8. An R-gene, Asc-1, from tomato plants of the asc/asc genotype, confers total resistance to AAL-toxin-induced programmed cell death9. Meanwhile, fusicoccin, a toxin produced by the fungus Fusicoccum amygdali, renders the stomata unable to close and ultimately causes plant wilting. It does so by irreversibly activating the plant’s plasma membrane H+-ATPase10. While these examples pertain to the involvement of toxins in total resistance, some phytotoxins are also involved in QDR.

Fungal tricothecenes play a role in QDR in wheat infected with Fusarium head blight. In this pathosystem, Fhb7 confers resistance to the disease by catalyzing the addition of glutathione (GSH) to a terminal epoxide of fungal tricothecenes11. A horizontally transferred glutathione S-transferase gene underlies this resistance mechanism conferred by Fhb7. Meanwhile, the HC toxin produced by the fungus Cochliobolus carbonum race 1 takes part in QDR in maize. The maize Hm1 gene encodes an HC toxin reductase that inactivates the HC toxin12. Although Hm1 confers major resistance, the partial loss-of-function of some Hm1 alleles confers QDR13.

Two other phytotoxins have been involved in QDR: the SS toxin in the Stemphylium solani-Allium sativum pathosystem and, in our laboratory, aldaulactone from Alternaria dauci, the causal agent of Alternaria Leaf Blight (ALB) on carrot (Daucus carota)14,15,16. The SS toxin causes plant cell death by inhibiting the H+-ATPase activity, NADH oxidation rate, and Fe(CN)63− reduction rate in a dose-dependent manner14. Thus, the SS toxin targets the standard redox system and the plasma membrane H+-ATPase, the latter mechanism being similar to that of fusicoccin. These findings bolster the long-standing notion that the plasma membrane is one of the primary sites of action of phytotoxins17. Meanwhile, aldaulactone is produced by the necrotrophic fungus Alternaria dauci, causing ALB, the most prevalent and damaging foliar disease in carrots18,19,20,21. This disease is characterized by necrotic lesions surrounded by a chlorotic halo on carrot leaves. Out of the over 70 phytotoxins known to be produced by the phytopathogenic species of Alternaria, six are known to be produced by A. dauci: zinniol, alternariol, alternariol monomethyl ether, α-acetylorcinol, p-hydroxybenzoic acid, and aldaulactone16,22,23,24,25,26. Moreover, the existence of 19 secondary metabolite gene clusters in its genome indicates that this list is most likely not exhaustive27. This quite diverse toxin weaponry perhaps helps A. dauci infect a wide range of dicotyledonous plants, mainly inside but also outside of the Apiaceae family, with the carrot being its main host28. There are various strains of A. dauci, each characterized by different levels of aggressiveness. Similarly, various genotypes of carrots exhibit different levels of resistance to A. dauci, broadening the disease resistance levels to a continuum29.

In recent years, we took advantage of the diversity of both carrot resistance level and A. dauci aggressiveness level to decipher the molecular basis of carrot QDR to the fungus. First, a correlation between plant partial resistance to A. dauci with in vitro cultured carrot cell resistance to fungal exudates was observed, indicating toxin resistance as a QDR mechanism15. Then, aldaulactone was isolated and characterized, dubbing it as an original A. dauci-produced phytotoxin. It likely plays an important role in the carrot-A. dauci interaction, as the production of aldaulactone was found to correlate with fungal pathogenicity levels16. In addition, in vitro assays showed that I2- partially resistant carrot cells were less susceptible to aldaulactone when compared to H1-susceptible carrot cells16. These results prompted us to better understand the structure of aldaulactone and its biosynthetic pathway.

Aldaulactone is a benzenediol lactone of a polyketide (PK) nature16. Generally, PK benzenediol lactones are comprised of two molecule families bearing a 1,3-benzenediol moiety connected to a macrolactone: either a dihydroxyphenylacetic acid lactone (DAL) or a resorcyclic acid lactone (RAL). Through bidimensional NMR analyses, aldaulactone was identified as a PK containing a DAL16. The biosynthesis of PKs entails polyketide synthase (PKS) genes that are usually organized in the same cluster. Tailoring enzymes that catalyze functional group transfer or redox reactions are also encoded by other genes found within the same cluster30. In fungi, two types of PKSs are known: type I and type III PKSs. While type III fungal PKSs are less characterized, type I fungal PKSs are widely studied since most fungal PKSs fall under this category31,32. Type I PKSs iteratively catalyze the head-to-tail Claisen condensation of acetyl-CoA, leading to the eventual formation of a polyketide30,33,34. These multifunctional enzymes can be grouped into three types, according to the resulting PK’s degree of reduction: non-reducing (NR)-, partially reducing (PR)-, and highly reducing (HR)-PKSs35,36. Moreover, PKSs are multidomain enzymes that harbor a minimal core module of three domains: ketosynthase (KS), acyltransferase (AT), and acyl carrier protein (ACP)37.

The biosynthesis of some RALs such as zearalenone and 10,11-dehydrocurvularin have been described, owing to their genetic and molecular underpinnings38,39,40,41. In these pathways, the molecular backbones of the toxins are synthesized by the joint effort of type I HR-PKS and NR-PKS. The same collaborative effort between the two type I PKSs is also observed in the synthesis of DALs42. The regioselectivity of the cyclization spells out the difference between the synthesis of a RAL from that of a DAL: a C2-C7 aldol condensation produces a RAL, while a C3-C8 aldol condensation yields a DAL42.

The construction of the biosynthetic pathway of aldaulactone requires probing into the genetic and molecular foundations of the PKSs involved. Bioinformatic analyses on the A. dauci genome revealed a single cluster (cluster 8) harboring two PKSs—AdPKS7 and AdPKS8—thought to be involved in aldaulactone biosynthesis27. The AdPKS7 encodes an HR-PKS, while the AdPKS8 encodes an NR-PKS. Based on the organization of both PKS genes in the cluster and the structure of aldaulactone, a biosynthetic pathway of this toxin was proposed27. The expression patterns of AdPKS7 and AdPKS8 correlated with aldaulactone production under different experimental conditions27. These findings support the hypothesis that both AdPKS7 and AdPKS8 are involved in aldaulactone biosynthesis.

Here, we present experiments designed to prove the implication of the two PKSs in aldaulactone biosynthesis and the involvement of aldaulactone in A. dauci pathogenicity. To our knowledge, no A. dauci transformation experiment has been published until now. The first aim of this study was thus to generate knock-out mutants of AdPKS7 and AdPKS8 in the FRA001 strain of A. dauci. Protoplast production, double-joint PCR-produced cassette uptake, and homologous recombination permitted the transformation of A. dauci. Two domains were targeted in each of the two AdPKS genes: the AT and the KS domains, which are minimally required to synthesize the aldaulactone backbone. In the mutants, the target domains were replaced by the Hygromycin Phosphotransferase gene (Hph), which confers resistance to hygromycin B A non-coding mutant (i.e., the target domain was outside of the coding region of both AdPKS genes) was also constructed. In total, five mutants were generated: AdPKS7∆AT, AdPKS7∆KS, AdPKS8∆AT, AdPKS8∆KS, and ∆NC.

The second aim of our study was to characterize the A. dauci mutants based on their ability to produce aldaulactone and their pathogenicity on carrot leaves. We analyzed the organic exudates from the five fungal mutants and the wild-type FRA001 strain through HPLC-UV. In a controlled environment, we also infected carrot leaves of both the H1-susceptible genotype and the I2-partially resistant genotype using conidial suspensions from all mutants and the wild-type. Our results indicate that the transformation method employed on A. dauci is efficient in carrying out functional validation experiments. Subsequently, through analyzing the mutants, we provided definitive proof of the function of AdPKS7 and AdPKS8 in aldaulactone production and the important role of this toxin in A. dauci pathogenicity on carrots.

Results

Transformation of A. dauci and molecular characterization of the transformed strains

The transformation of A. dauci was realized in several steps, which include the generation of the hygromycin B resistance cassette, protoplast production through enzymatic digestion, cassette DNA uptake in protoplasts, and selection of transformed fungal cells. The enzymatic cocktail Driselase/Kitalase allowed the A. dauci FRA001 strain to produce a high yield of protoplasts after six hours of digestion. The protoplast transformation protocol originally developed for A. brassicicola was adaptable to A. dauci, although differences in hygromycin sensitivity necessitated some modifications in selection concentration. The initial transformation to create A. dauci KO mutants was done using an agar overlay containing 12 µg·mL− 1 of hygromycin B, following the protocol for A. brassicicola43,44. However, no A. dauci transformants were recovered under this condition.

Subsequent hygromycin sensitivity assays with conidia from the A. dauci FRA001 and A. brassicicola Abra43 strains revealed that A. dauci is more sensitive to hygromycin B than A. brassicicola is, with complete inhibition of conidial germination at concentrations above 1 µg·mL− 1. A second transformation attempt was thus conducted with a reduced hygromycin concentration of 2 µg·mL− 1 to select KO mutants. Although successful, this second transformation revealed that 2 µg·mL− 1 hygromycin B was insufficient for effective selection, as both transformants and non-transformants traversed the hygromycin B-supplemented agar overlay. The transformants were then transplanted anew to PDA with either 5 or 8 µg·mL− 1 hygromycin B. This process determined that an agar overlay supplemented with 8 µg·mL− 1 of hygromycin B was effective for initial selection, with subsequent transfer to PDA supplemented with 5 µg·mL− 1 of hygromycin B to support sustained fungal colony growth without excessive inhibition.

The molecular characterization of the wild-type A. dauci FRA001 strain, along with all the mutants, was performed by PCR verification. Figure 1 shows the schematic diagram of the A. dauci secondary metabolism gene cluster 8 that harbors both AdPKS7 and AdPKS8, and the targeted regions replaced by Hph. Sequences corresponding to the internal transcribed spacer (ITS), hygromycin B phosphotransferase (Hph) gene, the acyltransferase (AT) and ketosynthase (KS) domains of AdPKS7 and AdPKS8, and the non-coding region targeted in the ΔNC strain were amplified via standard PCR and analyzed using gel electrophoresis. All A. dauci strains exhibited amplification of the ITS, whereas only the mutant strains displayed amplification of the Hph (Fig. 2). In the wild-type strain, the AT, KS, and non-coding regions were amplified. The targeted domains showed no amplification in the respective mutants.

To verify the correct insertion of Hph in place of the target domains, PCR amplification was performed across the region spanning the upstream sequence of the respective AdPKS gene (AdPKS7 or AdPKS8) and the downstream sequence of Hph. Similarly, amplification was carried out for the downstream region of the AdPKS gene and the upstream region of Hph. The presence of PCR products corresponding to both regions in the gel confirmed the correct insertion of Hph into the genomes of the A. dauci mutants (Fig. 2). These results corroborate the successful knockout of the PKS genes, replaced by Hph in the PKS mutants, as well as the substitution of the non-coding region with Hph in the ΔNC mutant.

Fig. 1
figure 1

Schematic representation of the Alternaria dauci secondary metabolism gene cluster 8 (GenBank number: PP445245.1), showing the arrangement and orientation of predicted genes, as well as the areas targeted for replacement with Hph. Green blocks denote genes with putative functions, while black blocks indicate uncharacterized open reading frames. The two iterative PKSs, AdPKS7 and AdPKS8, are depicted as white blocks containing colored domains that were replaced by Hph: the cyan ΔAT and the orange ΔKS in both the AdPKS7 and AdPKS8, marking four targeted knockout regions. The fifth targeted region, the ΔNC downstream of the AdPKS7, is shown in yellow. The scale corresponds to genomic distance in kilobases.

Full size image
Fig. 2
figure 2

Molecular characterization of A. dauci strains through PCR verification, visualized using 1.2% agarose gels. The individual gel photos are numbered 1 to 17 from top to bottom. Gel 1, amplification of ITS; gel 2, Hph; gel 3, AT domain of AdPKS7; gel 4, KS domain of AdPKS7; gel 5, AT domain of AdPKS8; gel 6, KS domain of AdPKS8; gel 7, NC region; gel 8, upstream region of AdPKS7ΔAT and downstream region of Hph; gel 9, downstream region of AdPKS7ΔAT and upstream region of Hph; gel 10, upstream region of AdPKS7ΔKS and downstream region of Hph; gel 11, downstream region of AdPKS7ΔKS and upstream region of Hph; gel 12, upstream region of AdPKS8ΔAT and downstream region of Hph; gel 13, downstream region of AdPKS8ΔAT and upstream region of Hph; gel 14, upstream region of AdPKS8ΔKS and downstream region of Hph; gel 15, downstream region of AdPKS8ΔKS and upstream region of Hph; gel 16, upstream region of ΔNC and downstream region of Hph; gel 17, downstream region of ΔNC and upstream region of Hph. Lane 1, wild-type A. dauci FRA001 strain; Lane 2, AdPKS7ΔAT; lane 3, AdPKS7ΔKS; lane 4, AdPKS8ΔAT; lane 5, AdPKS8ΔKS; lane 6, ΔNC; lane 7, negative control. While the gels in this figure are cropped, the individual, whole gel photos are provided in Supplementary document 1.

Full size image

Morphological profiling of A. dauci mutants

To our knowledge, this study presents the first reported case of genetic transformation in A. dauci. Consequently, we documented the morphological characteristics of the resulting mutants, focusing specifically on their radial mycelial growth on V8 agar medium, mycelial morphology, and conidial morphology.

The mycelial growth rates of all six A. dauci strains were evaluated by computing the Area Under the Growth Curve (AUC) values. Figure 3a presents a statistical comparison of the mean AUC values across nine replicates per strain, analyzed using Waller-Duncan post-hoc test. The mycelial growth rates of all five mutant strains were slightly yet significantly faster than that of the wild-type. Moreover, the ΔNC mutant displayed a growth rate that was statistically similar to those of the PKS mutants. A visual comparison of mycelial growths on V8 agar medium at seven days post-transplant is shown in Fig. 3b. While differences in growth rate between the wild-type and mutant strains were subtle, a marked contrast in mycelial pigmentation was evident: the wild-type strain displayed a lighter hue compared to the black pigmentation observed in all mutant strains. No difference in mycelial morphology was observed (Supplementary document 2).

Conidial phenotyping of the mutants was also conducted to evaluate further pleiotropic effects of transformation. The number of longitudinal septa in conidia varied significantly among the A. dauci strains, as revealed by Kruskal-Wallis analysis followed by pairwise Wilcoxon tests (p < 0.05). The wild-type FRA001 strain exhibited the lowest number of longitudinal septa, differing significantly from all five mutants. The AdPKS8ΔAT mutant showed the highest median septation. Letter groupings assigned based on post-hoc comparisons indicate that all mutant strains had significantly more longitudinal septa than the wild-type (Fig. 4a).

Figure 4b compares representative conidia from each A. dauci strain, illustrating the noticeably simpler internal structure of the wild-type conidium compared to those of the mutant strains. Under typical conditions, A. dauci conidia display an ellipsoidal body with multiple transverse septa and one or more longitudinal septa, tapering into a long, slender filiform beak (Fig. 4b, FRA001). Conidia from the mutant strains displayed increased longitudinal septation, and in some cases, abnormal morphologies, such as bent bodies often with a clump of cells, as seen in the micrograph of the AdPKS7ΔKS conidium.

Fig. 3
figure 3

Mycelial growth phenotyping analysis. (a) Statistical comparison of average Area Under the Growth Curve (AUC) values calculated from the radial mycelial growth of A. dauci strains on V8 agar medium. Three biological repetitions of three replicates per fungal strain were obtained and then analyzed using Waller-Duncan test. Different letters indicate statistically significant differences (p < 0.05). (b) Obverse view of mycelial growth of A. dauci strains on Petri dishes containing V8 agar medium. Photographs were taken seven days post-transplantation of the mycelial plug into the center of the plate.

Full size image
Fig. 4
figure 4

Conidial phenotyping analysis. (a) Statistical comparison of mean longitudinal septa number derived from 35 conidia per A. dauci strain using Kruskal-Wallis test, followed by pairwise Wilcoxon tests. Different letters indicate statistically significant differences (p < 0.05). (b) Micrographs of representative A. dauci conidia for each strain, illustrating morphological differences. Scale bar = 100 μm.

Full size image

HPLC-UV analysis for aldaulactone detection

The organic exudates (OEs) of the various A. dauci strains were analyzed using HPLC-UV to assess their ability to produce aldaulactone. For each fungal strain, three biological replicates were included. Since aldaulactone has a maximum absorbance at 305 nm, all chromatograms were examined at this wavelength16. Aldaulactone exhibits a retention time of 18.5 min, as seen in the chromatogram of the aldaulactone standard (i.e., pure aldaulactone; Fig. 5). A peak corresponding to aldaulactone was detected in the wild-type and ΔNC strains. However, this peak was absent in all ΔPKS mutants.

To confirm the identity of the purported aldaulactone detected in the wild-type and ΔNC strains, the UV profiles of the compounds from the 18.5-min peak in each chromatogram were compared with that of aldaulactone. The profiles were consistent across three biological replicates, showing characteristic absorbance peaks at 196 nm and 302 nm (Supplementary document 3). These results confirm that the peak with a retention time of 18.5 min observed in both the wild-type and ΔNC strains corresponds to aldaulactone.

Fig. 5
figure 5

HPLC-UV absorption profiles of the organic exudates from various A. dauci strains analyzed at 305 nm, in comparison with the aldaulactone standard. The line positioned at the retention time of 18.5 min indicates the aldaulactone peak. For all fungal strains, peak intensities were adjusted according to the ratio of mAU to the actual volume of raw organic extract injected into the HPLC column. Before launching, all organic exudates were adjusted to a final concentration of 5 mg·mL− 1. Supplementary document 3 presents the corresponding UV profiles showing the absorbance of compounds isolated at 18.5 min for all A. dauci strains.

Full size image

Carrot in planta pathogenicity test

To assess the ability of the mutant strains to induce symptoms characteristic of Alternaria Leaf Blight (ALB), we inoculated carrot leaves with conidial suspensions from both the wild-type and mutant A. dauci strains. The degree of pathogenicity of the A. dauci strains on carrot leaves was then compared by analyzing the log-transformed Area Under the Disease Progression Curve (LogAUDPC) values (raw data available in Supplementary document 4). An example of symptom severity on carrot leaves showing different LogAUDPC values is provided in Fig. 6a.

A Waller-Duncan post-hoc analysis was conducted to statistically compare between symptom severity of each A. dauci strain. Given that the ΔNC strain shares closer genetic similarity to the ΔPKS mutants due to the insertion of the Hph gene, and exhibits similar morphological characteristics, we deemed it more appropriate to compare the pathogenicity of the ΔPKS mutants to the ΔNC strain rather than the wild-type FRA001 strain. In the H1-susceptible carrot genotype, the ΔNC strain exhibited significantly higher pathogenicity compared to the ΔPKS mutants, though no significant difference was observed relative to the wild-type strain (p < 0.05; Fig. 6b). For the wild-type FRA001 strain, a significant difference in disease severity was noted between the H1-susceptible and I2-partially resistant carrot genotypes, with the former showing a markedly higher level of symptom severity. A similar pattern was observed for the ΔNC strain, where the H1-susceptible genotype displayed significantly higher disease severity than the I2-partially resistant genotype. In contrast, no significant difference was seen between the H1-susceptible and I2-partially resistant leaves infected with any of the ΔPKS mutants. The identity of the A. dauci used for inoculation was confirmed by microscopic viewing of the conidia still attached to the carrot leaves 48 h following leaf detachment (Supplementary document 5).

Fig. 6
figure 6

In planta pathogenicity test results. A. dauci FRA001 and mutant strains were used to inoculate the H1-susceptible and I2-partially resistant cultivars. The experiment was repeated four times, and each time, three technical replicates were used, resulting in 12 data points per cultivar-strain combination. (a) Statistical comparison of LogAUDPC values among various A. dauci strains, as well as between the H1-susceptible and I2-partially resistant carrot genotypes. The values at the top indicate the p-values resulting from comparisons of each A. dauci strain against the ΔNC strain within the H1-susceptible carrot genetic background. The values at the bottom represent p-values obtained from comparisons between the two carrot genotypes infected with the specified A. dauci strain. Data were analyzed using the Waller-Duncan post-hoc test (p < 0.05), with values below 0.05 indicating significant differences. (b) Carrot leaves with varying LogAUDPC values, showing different levels of disease severity. Before photographing, the inoculated third-stage leaves of six-week-old carrots were detached from the whole plant, placed in a Petri dish lined with moist filter paper, and stored at 4 °C for two days.

Full size image

Discussion

As a necrotrophic pathogen, A. dauci is expected to rely on phytotoxins to establish pathogenicity in its carrot host. To date, six secondary metabolites (SMs) with a phytotoxic activity have been identified as being produced by A. dauci16,22,23,24,25,26. While the role of zinniol as a phytotoxin has been questioned, aldaulactone accounts for most—but not all—of the in vitro toxicity of A. dauci exudates to carrot cells16,45. In addition, aldaulactone production is strongly correlated with A. dauci pathogenicity. Finally, a notable correlation was observed between carrot cell resistance to A. dauci organic extracts and carrot plant resistance to the fungus, which led us to think that toxin resistance is a major resistance mechanism in the carrot-A. dauci interaction15,16.

To obtain a snapshot of the genetic determinants of secondary metabolism in A. dauci, our lab has provided the first transcriptome of this species, along with the first appraisal of SM core gene diversity within Alternaria genomes27. These investigations pinpointed AdPKS7 and AdPKS8, both located in gene cluster 8, as candidate genes responsible for aldaulactone biosynthesis27. To confirm this, we conducted a functional validation of these genes that prompted us to generate and analyze KO mutants. To accomplish this goal, we adapted a transformation method originally developed for A. brassicicola Abra4343,44. The method was successfully applied to the A. dauci FRA001 strain with a couple of modifications. First was the use of 8 µg·mL− 1 hygromycin in the agar overlay, compared to 12 µg·mL− 1 used for the transformation of A. brassicicola Abra43. This adjustment was necessary because A. dauci transformants exhibit lower resistance to hygromycin B than A. brassicicola Abra43 transformants. Another modification involved extending the enzymatic digestion time for A. dauci FRA001. While A. brassicicola Abra43 cells underwent digestion for four hours, those of A. dauci FRA001 were treated for six hours. To our knowledge, this is the first reported case of targeted genetic manipulation in A. dauci.

Morphological profiling of the transformed A. dauci strains revealed phenotypic effects associated with its genetic transformation. Significantly faster mycelial growth and larger, more septate conidia were observed in all mutant strains including the ΔNC strain, compared to the wild-type. This suggests that the transformation process itself impacted their physiology. Alternatively, gene insertion and/or disruption within cluster 8 has widespread effects on fungal phenotype whether one of the genes in this cluster is disrupted or not. Fungal transformation experiments have had the reputation of fostering mutational events, but a recent study based on genome resequencing indicates that – at least in the case of Cryptococcus neoformans – mutation rate is not high among transformed fungal strains46. Besides, mutations are random events while the phenotypic variations we observed are consistent among all our transformed strains. These effects could be due to the insertion of foreign DNA, such as the Hph gene, which may have activated a general stress response pathway, potentially leading to physiological reprogramming. To our knowledge, the insertion of the Hph gene has not been associated with similar phenotypic changes in other fungal species. However, a study on the development of spontaneous hygromycin B resistance in the necrotrophic pathogen Monilinia fructicola demonstrated that hygromycin B-resistant mutant colonies exhibited slower growth on PDA medium when compared to the wild-type47. Although this study did not employ site-specific mutagenesis, it suggests that the insertion of the Hph gene might lead to developmental changes in fungi. Interestingly, similar conidial form variations have also been observed in other A. dauci strains, such as FRA017, a wild-type strain resistant to the fungicide, iprodione. Also, similar morphological variations were observed in A. dauci conidia exposed to environmental stress, such as fungicide48. These findings suggest that the observed morphological changes can correspond to the expression of stress resistance-associated mechanisms. Another hypothesis is that the protoplast stages during the transformation, as well as the monosporing stages, selected more mitotically active transformants, explaining both the faster mycelial growth of the fungus and the larger, more septate conidia observed. The shared morphological features observed among the ΔPKS mutants and the ΔNC strain, coupled with the insertion of Hph into their genomes, led us to use the ΔNC strain rather than the wild-type as the basis of comparison in experiments assessing the effects of AdPKS gene knockouts.

Organic exudates from the A. dauci strains were analyzed by HPLC equipped with a photodiode array detector. Chromatograms were extracted at 305 nm, the wavelength where aldaulactone shows maximum absorbance. The ΔPKS mutants did not show the characteristic aldaulactone peak at 18.5 min. In contrast, both the wild-type and the ΔNC strains exhibited the aldaulactone peak, which is consistent across all three biological replicates. Moreover, the UV profile of the compound detected at 18.5 min in these two strains closely matches that of aldaulactone (Supplementary document 3). Aldaulactone accumulation seemed somewhat weaker in the ΔNC strain than in the wild-type strain. It can be hypothesized that the insertion of the Hph gene within the gene cluster in the ΔNC quantitatively affected AdPKS7 and AdPKS8 expression and aldaulactone production. Minor peaks were observed beyond the 18.5 min retention time across all fungal strains, suggesting that other metabolic pathways involved in the production of the compounds in these peaks were not affected by the transformation process. However, minor peaks detected before 18.5 min were present in the wild-type FRA001 strain but not in the transformed strains. Other metabolic pathways involved in the production of the compounds detected in these peaks may have been affected by the transformation process and not the knock-out of either of the AdPKS genes, as the ΔNC strain did not show similar peaks. Since AdPKS7 and AdPKS8 are close together within cluster 8, it can be hypothesized that the disruption of one of these two genes affects the expression level of the other. According to the model proposed by Courtial et al. (2022), aldaulactone backbone synthesis proceeds sequentially: AdPKS7 synthetizes a triketide precursor, which is subsequently transferred to AdPKS8, which adds four more malonyl-coA units and performs cyclization of the aldaulactone backbone27. In this model, the knockout of either AdPKS gene results in the impossibility for the fungus to produce aldaulactone, regardless of the expression level of the other AdPKS gene. The results we obtained are coherent with this model. All in all, these results provide strong evidence that AdPKS7 and AdPKS8 are responsible for aldaulactone biosynthesis. In addition, they provide us with a set of isogenic strains that only differs in their production of aldaulactone.

To evaluate the role of aldaulactone production in fungal pathogenicity, we conducted an in planta pathogenicity test on six-week-old carrot leaves. In the H1-susceptible carrot genotype, a significant difference in disease severity was observed between the ΔNC strain and all ΔPKS mutants, but not between the ΔNC strain and the wild-type strains. Indeed, the non-significant difference between the wild-type FRA001 and the ΔNC strain does not suggest any form of pathogenicity loss. This suggests that aldaulactone plays an essential role in the pathogenicity of A. dauci. In contrast, no significant difference was observed across all fungal strains when inoculated on the I2-partially resistant carrot genotype. Aldaulactone production is the only component of pathogenicity that differs between the ΔNC and the ΔPKS mutants. In a previous study, we saw that the I2 cultivar resists aldaulactone toxicity16. A possible explanation for the in planta observation could be that since the I2 genotype resists aldaulactone, it is solely the other components of A. dauci pathogenicity that are responsible for the symptoms observed in this genotype. As a result, the absence or presence of aldaulactone alone does not affect the outcome of the confrontation with A. dauci strains unable to produce aldaulactone, suggesting that the other components of I2 resistance are not strong enough to be detected in our set-up when compared with that of the H1 genotype.

Our study demonstrates that the knockout of AdPKS7 or AdPKS8, resulting in the complete loss of aldaulactone production, significantly reduces disease symptom severity in the H1-susceptible carrot genotype. This finding strongly indicates that aldaulactone is the primary phytotoxin responsible for causing ALB symptoms in H1-susceptible carrots. These results are in contrast with a recent study on the functional redundancy of pathogenicity factors in Botrytis cinerea. In that study, multiple knockout mutants, lacking up to 12 cell-death-inducing proteins (CDIPs) and metabolites, exhibited a progressive decrease in virulence49. Interestingly, the reduction in virulence correlated with the increasing number of gene knockouts, and was not significant when less than four genes were deleted, suggesting functional redundancy among the pathogenicity factors. Despite the substantial reduction in known phytotoxins, significant phytotoxic activity persisted in the mutants’ secretomes, pointing to the presence of yet-unidentified CDIPs in B. cinerea. In stark contrast, our findings show that the knockout of AdPKS7 or AdPKS8 leads to a clear reduction in ALB symptoms in carrots. Unlike B. cinerea, A. dauci does not exhibit compensatory mechanisms to recover its pathogenicity when the aldaulactone biosynthesis pathway is disrupted. This underscores the central role of aldaulactone in A. dauci pathogenicity on carrots.

Our previous findings have established that aldaulactone is toxic to carrot cells and also to N. benthamiana leaves when directly injected into them, suggesting that aldaulactone may be classified as a non-host-specific toxin (NHST)27. These findings are in line with previous experiments showing that A. dauci is pathogenic to many dicotyledonous plants besides carrot, including N. benthamiana, corn salad, cress, radish, and tomato27,28. This broader host range suggests that aldaulactone potentially targets conserved cellular processes, a hallmark of NHSTs. The resistance mechanisms against NHSTs, such as aldaulactone, are often associated with QDR rather than race-specific total resistance typical of HSTs. In this context, the partial resistance observed in the carrot I2 genotype likely demonstrates an array of possible defense responses, such as detoxification mechanisms and cellular defense mechanisms, among others. This is in contrast with the HC toxin resistance mechanism in maize, where the single dominant HM1 gene encodes a specific HM toxin reductase that inactivates the toxin. The NHST classification of aldaulactone thus corresponds to the general notion of this group of toxins in that they are not the sole determinant of pathogenicity50. In turn, the ALB symptoms observed on carrot leaves infected with the ΔPKS mutants are attributed to the other components of A. dauci pathogenicity and not aldaulactone.

Both the ingenuity and downfall of toxin resistance mechanisms in plants are reflected in the plants’ evolutionary race against necrotrophic pathogens. For instance, the Pc2 gene renders some plants susceptible to victorin, a toxin produced by the necrotrophic fungus, Cochliobolus victoriae51. Various Tsn genes, on the other hand, recognize effectors encoded by Tox genes in Parastagonospora nodorum52. Upon recognition of the effectors, the Tsn genes activate programmed cell death, a means for necrotrophic fungi such as P. nodorum to undermine plant defense responses and promote infection. These are examples of the inverse gene-for-gene hypothesis, where a susceptible gene, instead of an R-gene, is compatible with the gene of a pathogen encoding an avirulence factor. Another example is the Asc1 gene in tomatoes that confers susceptibility to AAL-toxin produced by Alternaria alternata f. sp. lycopersici53. The mutation in Asc1 in some tomato cultivars is responsible for the susceptibility to the toxin. These cases show that plants resist fungal-derived phytotoxins in various ways, whether by detoxifying the toxins or by lacking susceptibility genes. However, how carrots resist aldaulactone is yet to be determined. As mentioned, the NSHT nature of aldaulactone suggests that resistance against it may not depend on a single R-gene, but rather on general cellular defense mechanisms or perhaps, on the inverse gene-for-gene hypothesis.

All in all, we provided an efficient site-directed transformation method for A. dauci, offering a valuable tool for genetic manipulation in this species. Through the analysis of the transformed strains, definitive proof of the roles of AdPKS7 and AdPKS8 in aldaulactone production was provided. Furthermore, fungal transformation gave us isogenic lines that helped us isolate the role of aldaulactone in both fungal pathogenicity and carrot plant resistance. This study represents the first demonstration of the role of both aldaulactone itself and aldaulactone resistance in the carrot-A. dauci interaction. Further research is needed to elucidate the toxin mode of action and the resistance mechanisms involved in this pathosystem. That is, studies streamlined on identifying the molecular target of aldaulactone and the key defense pathways it activates could reveal whether the resistance against it fits into known mechanisms or follows an entirely different one. Beyond the recessive resistance to necrotrophic effectors such as those encoded by P. nodorum Tox genes, there are only three examples where pathogen resistance genes were shown to encode toxin resistance factors11,12,53,54. Lengthening this short list would invite insights into the diversity of plant toxin resistance mechanisms, a crucial part of plant resistance mechanisms against necrotrophic pathogens.

Methods

Fungal material and fungal growth conditions

Alternaria dauci was grown on Petri dishes containing V8 agar and incubated at 24 °C in darkness for 14 d. Only one strain of A. dauci was used in this paper: the FRA001 strain, which is moderately aggressive on carrot and was also used in other studies16,27,28. It was collected as described in28,55. Alternaria brassicicola strain Abra43 was collected as described in55. FRA001 and Abra43 resistance to hygromycin was assessed by transferring conidia on PDA medium plates containing either no hygromycin, 0.5, 1, 2.5, 5–10 µg·mL− 1 hygromycin. Both strains are freely available from the COMIC collection (COMIC-SFR QuaSaV, 42 rue Georges Morel, 49070 Beaucouzé cedex, France).

For liquid cultures, all fungal strains were grown in Difco™ Potato Dextrose Broth (PDB) as in16. Ten colonized agar plugs (about 5 mm in diameter) were taken from V8 agar culture and ground for 1 min in 40 mL of PDB. Sixty mL of PDB was then used to rinse any adhering mycelia on the sterile metal grinder (DuPont Instruments, Sorvall® Omni-Mixer), then poured into the flask containing the earlier collected 40-mL suspension. The 100-mL suspension was then sealed and incubated in the dark at 24 °C for 72 h with an agitation of 125 rpm (MINITRON®, Infors HT).

HPLC-UV analysis

A. dauci FRA001 PDB culture suspensions for HPLC-UV analysis were grown for 72 h. Then, the raw exudates were recovered by serial filtration using 200 μm, 50 μm, and 1 μm sterile nylon membranes from Sefar Nitex (Sefar AG, Heiden, Switzerland) as in15. The mycelia were weighed and stored at -80 °C for future use, while the filtered raw exudates were neutralized to pH 7 and underwent liquid-liquid extraction through the addition of an equal volume of ethyl acetate. This was done three times to ensure that the organic exudate (OE) was maximally extracted. The OEs were then dried over sodium sulphate and evaporated under reduced pressure, and stored as described in15.

The OEs were resuspended in HPLC-grade methanol (5 mg·mL− 1) and centrifuged at 11,000 g for 10 min. The supernatants were analyzed using a Shimadzu Prominence-I LC-2030 3D coupled with PDA detector and assisted by LabSolutions software. The mobile phase (flow rate: 0.7 mL·min− 1) was composed of water and HPLC-grade methanol with the following gradient: 90% water and 10% methanol at 0 min, then 100% methanol after 25 min, maintained for 5 min. All analyses were performed at 25 °C on an UPTISPHERE C18-ODB column (150 × 4.6 mm; 3 μm).

Genomic DNA extraction and standard PCR

The genomic DNA of A. dauci was extracted using the microwave miniprep method as described by56. The molecular characterization of fungal strains through standard PCR (Thermocycler T100, Bio-rad) was performed on various genomic regions, with the internal transcribed spacer (ITS) region as a positive control. The PCR mixtures contained 1x Green GoTaq® Flexi Buffer (Promega), 1.25u of GoTaq® Hot Start DNA polymerase (Promega), 0.1 µM of each primer, 1.5 mM of MgCl2 solution, 0.2 mM of dNTP, 1 µL of the DNA suspension, and diluted to a total volume of 25 µL using nuclease-free water. The primers used in this study are listed and detailed in Supplementary document 6. The PCR programs used vary depending on the regions amplified. Detailed programs are specified in Supplementary document 7.

Gene replacement cassette production and fungal transformation

The gene replacement cassettes were created by the double-joint PCR method as detailed by57. They bore the 5’ and 3’ flanking regions of each target sequence fused with the Hph gene (GenBank number: LT726869) conferring resistance to hygromycin B. The target sequences were defined using the sequence of A. dauci secondary metabolism cluster 8 (GenBank number: PP445245.1). For each amplification, the Phusion Hot Start II High-Fidelity DNA polymerase pack (Thermo Fisher Scientific) and the appropriate primers were used. The final volume for each reaction was 50 µL, which was comprised of Phusion Hot Start DNA polymerase (0.02 u·µL− 1), 3% v/v DMSO, 1x Buffer HF, dNTP (0.2 mM each), primers (20 µmol each), and 1 µL of the DNA. The first round of PCR amplified the 5’ and 3’ flanking regions of the target gene, as well as the Hph gene (1699 bp) from the plasmid pCB163658. The second round of PCR ligated the products from the first PCR round to form the gene replacement cassette. The third round of PCR amplified the entire cassette from the second PCR round. Products of the first and third rounds of PCR were purified using the NucleoSpin® Gel and PCR Clean-up kit (Macherey-Nagel) according to the manufacturer’s instructions.

Protoplast production was initiated by culturing A. dauci conidia (minimum of 80 cfu·mL− 1) in 100 mL of PDB for 15 h at 24 °C with an agitation of 175 rpm. The mycelia were then collected by centrifugation (10 min, 1700 g), washed in 0.7 M NaCl, and then centrifuged again using the same conditions. This washing step was performed twice before adding the mycelia to a 20-mL digestion solution containing Driselase (20 mg·mL− 1; Sigma D9515-5G) and Kitalase (10 mg·mL− 1; Wako 1W114-0037) prepared in 0.7 M NaCl. The resulting mixture was carefully mixed by inversion every 30 min for 6 h. The protoplasts were collected by centrifugation (7 min, 1700 g) and then delicately resuspended in 10 mL STC buffer (1.2 M Sorbitol, 10 mM Tris pH 7.5, 50 mM CaCl2). Centrifugation was repeated and the protoplasts were resuspended in 1 mL STC at a concentration between 106 and 108 protoplasts·mL− 1. For each mutant, 10–20 ng of the gene replacement cassette was added to 500 µL of protoplast suspension, which was then incubated on ice for 20 min. PEG [6% w/v Polyethylene glycol MW 3350 (Sigma-Aldrich P3640), 10 mM Tris-HCl pH 7.5, 50 mM CaCl2] was added three times, with the suspension warmed in the hand in between each addition. The suspension was gently mixed, incubated on ice for 5 min, and added with 1 mL of STC. Then, 250 µL of the protoplast suspension was added to 20 mL of the regeneration medium (1 M sucrose, 0.1% w/v yeast extract, 0.1% w/v casein hydrolase, and 1.6% w/v bacteriological agar), and the resulting mixture was poured into a Petri dish. After 24 h, a 10-mL agar overlay supplemented with hygromycin B (8 µg·mL− 1) was poured. The plates were stored at 24 °C in the dark.

The hygromycin B-resistant mutants were identified based on their ability to cross the agar overlay five to seven days after. They were then sub-cultured on PDA media supplemented with hygromycin B (5 µg·mL− 1). After a first round of monosporing, the purity of the mutants was checked by standard PCR as described above. In cases where the mutants were not pure (i.e., when a band corresponding to the target gene appeared on the agarose gel), monosporing on PDA media supplemented with hygromycin B (5 µg·mL− 1) was repeated. The mutant strains were conserved in cryotubes with 1 mL 30% glycerol and stored at -80 °C. In total, five A. dauci mutants were constructed from the A. dauci FRA001 strain: AdPKS7ΔAT, AdPKS7ΔKS, AdPKS8ΔAT, AdPKS8ΔKS, and ΔNC. The ΔNC strain was generated by replacing a non-coding region 946 bp downstream of the AdPKS7 gene with the Hph resistance cassette.

Fungal phenotyping

The mycelial radial growth assay was carried out according to59. Briefly, agar disks were sourced from the margin of a 7d-old colony grown on V8 agar media and were transferred to another V8 agar media. The radius of the mycelial colony was measured in mm at 3-, 4-, 5-, 6-, and 7 days post-transfer. The area under the growth curve was then calculated as in60 and photos of the obverse view of the Petri dishes bearing the mycelia were captured on the final day of measurement.

Conidial phenotyping was performed by first scraping the surface of 14-day-old fungal colonies from V8 agar cultures with 100 mL 0.05% Tween 20 solution. Three µL of the conidial suspension was mounted on a slide and then viewed under a light microscope (ZEISS Axio Imager 2) at 40x magnification. Mycelial phenotyping was performed using the same microscopy techniques on small samples of mycelium-colonized V8 agar medium, taken from the periphery of seven-day-old colony.

Plant material and in planta pathogenicity test

The Daucus carota genotypes used in this study were H1 (susceptible) and I2 (partially resistant). The H1 and I2 seeds were obtained as described in15. Plant cultivation was performed according to21. Briefly, plants were grown in pots containing peat moss/sand mixture in greenhouse conditions (16 h of day, 22 °C day/19°C night, 60% humidity) for six weeks.

Plant inoculation procedures have been described in detail in15,19. Sporulation of A. dauci strains was realized by individually growing the fungi on V8 agar media and incubating them at 24 °C in darkness for 14 d. The conidial suspensions were adjusted to 200 conidia mL− 1 in 0.05% Tween 20. The third leaf of six-week-old carrot plants was inoculated with the conidial suspension. For this, leaves still-attached to the plant were placed in incubation chambers made of a Petri dish anchored to the substrate. In an incubation chamber, a moist filter paper was placed at the bottom, allowing the leaf to rest on top of the paper and be held in place by paper clips and 6 mm nuts. Forty drops of 5 µL of the conidial suspension were applied on the adaxial side of the leaf using a micropipette. Symptom intensity was evaluated at 7-, 11-, and 13-days post-inoculation and is expressed as the number of symptoms per conidia. A value of 1 was assigned to a necrotic lesion smaller than the drop deposited, while a value of 10 was assigned to the lesion larger than the drop. The ratio of the number of lesions per viable conidia was calculated at each scoring date, and the Area Under the Disease Progression Curve (AUDPC) was determined according to19. In each condition (i.e., carrot genotype inoculated with the fungal strain), three technical replicates were made. This was done in four repetitions.

Statistical analyses

All statistical analyses were performed in RStudio (version 2025.05.0 + 496)61. Normality and homoscedasticity of residuals were checked using a Shapiro-Wilk test and a Levene’s test, respectively. For comparisons of mycelial radial growths, one-way analysis of variance (ANOVA; α = 0.05) was conducted, followed by a Waller-Duncan post-hoc test. Each fungal strain was represented by three independent biological replicates. The entire experiment was repeated three times.

For the analysis of longitudinal septa number, 35 random conidia were used. A Kruskal-Wallis test (α = 0.05) was performed, due to violations of ANOVA assumptions: the Shapiro-Wilk test indicated non-normality (p = 0.0054) and the Levene’s test showed heterogeneity of variances (p = 0.0139). The Kruskal-Wallis test was followed by pairwise Wilcoxon tests to identify significantly different strains (α = 0.05).

For the in planta pathogenicity test, AUDPC values were log-transformed (LogAUDPC) and analyzed by one-wayANOVA (α = 0.05) followed by Waller-Duncan multiple comparisons tests. For each condition (i.e., each carrot genotype-fungal strain combination), four biological repetitions of three technical replicates were performed.