Defects in the Ferroxidase That Participates in the Reductive Iron Assimilation System Results in Hypervirulence in Botrytis Cinerea

The plant-pathogenic fungus B. cinerea causes enormous economic losses, estimated at anywhere between $10 billion and $100 billion worldwide, under both pre- and postharvest conditions. Here, we present the characterization of a loss-of-function mutant in a component involved in iron acquisition that displays hypervirulence. While in different microbial systems iron uptake mechanisms appear to be critical to achieve full pathogenic potential, we found that the absence of the ferroxidase that is part of the reductive iron assimilation system leads to hypervirulence in this fungus. This is an unusual and rather underrepresented phenotype, which can be modulated by iron levels in the plant and provides an unexpected link between iron acquisition, reactive oxygen species (ROS) production, and pathogenesis in the Botrytis-plant interaction.

FET3 ferroxidase and the FTR1 permease that form the RIA system in this fungus (37) is required for virulence. This observation sharply differs from what has been described for C. heterostrophus, in which mutants lacking any of the RIA components do not display virulence alterations (38). As in the case of the NRPS-dependent iron uptake system that has not been studied in detail in B. cinerea, RIA has not been examined either. This highlights that in this particular phytopathogen, the impact of iron acquisition on virulence is unknown (39).
Although it has been suggested that iron content should be considered a strong environmental cue that modifies plant-pathogen interactions from the host-defensive perspective (20), iron uptake and traffic may also be important for the pathogen. Because very little is known about how B. cinerea obtains iron during infection, and considering that any significant perturbation in iron homeostasis will impact both plant immunity and pathogen virulence, we altered, by genetic disruption, the B. cinerea RIA system. Unexpectedly, although the ferroxidase (referred to here as BcFET1) mutant of the RIA system resulted in an iron-dependent phenotype, exhibiting significantly reduced whole-cell iron content during saprophytic growth, it was also found to exhibit augmented virulence. In contrast, the mutant for the iron permease (BcFTR1) did not reveal major iron-dependent phenotypes and displayed unaffected virulence. Our results suggest a role for BcFET1 in modulating Fe-dependent ROS generation and the effect of such regulation in impacting virulence.

RESULTS
Identification of an RIA-related ferroxidase-encoding gene (bcfet1) among multicopper oxidases of B. cinerea. To explore the role of the reductive iron assimilation (RIA) system in B. cinerea infection dynamics, we searched for genes that encode orthologues of the FET3/FTR1 system in its genome database. As mentioned, both proteins constitute the canonical fungal iron uptake complex (34). Employing the FET3 protein sequence from S. cerevisiae as bait, BLAST analysis retrieved 13 protein sequences containing multicopper oxidase (MCO) domains. The first hit identified was bclcc13 (described below). Therefore, we further analyzed the identified MCO sequences to distinguish ferroxidases among all identified proteins.
As indicated in Table S1 in the supplemental material, the identified MCO sequences possess annotated names that range from BcLcc1 to BcLcc13. A single MCO sequence was additionally identified when BcLcc1 to BcLcc13 were used as a BLAST analysis query (Bcin08g03600). The initials "Lcc" commonly refer to laccase enzymes, which are phenol oxidases that belong to the large family of MCO proteins described in different organisms, including plants and fungi (35,36). Nevertheless, although similar, they are distinct from ferroxidases. In the case of B. cinerea, only two laccases have been studied in detail, BcLcc1 and BcLcc2, which were first described in this fungus several years ago (40). Hence, all the mentioned proteins were inspected using multiple-sequence alignments. First, we analyzed previously defined laccase signatures (41) that differ in MCOs with ferroxidase activity, such as FET3. As shown in Fig. S1, the L3 signature is conserved across all analyzed sequences. In contrast, the L1 signature differs in the cases of BcLcc6, BcLcc7, BcLcc8, BcLcc11, and BcLcc13, while L2 and L4 signatures were more conserved among BcLcc sequences ranging from BcLcc1 to BcLcc12. Notably, for BcLcc13, all signatures were more similar to those of the S. cerevisiae FET3 protein, making this protein a suitable RIA-related ferroxidase (and not a laccase) candidate for further inspection.
Accordingly, a phylogenetic analysis employing the one-click method of the Phylogeny.fr platform was conducted (42). As observed in Fig. 1a, the phylogenetic reconstruction shows two major clades. In the first clade, all laccase MCO enzymes (BcLccs) of B. cinerea were clustered. In contrast, in the second, only BcLcc13/ Bcin02g02780 (here termed BcFET1) grouped with the analyzed RIA-related ferroxidases, including those from different filamentous fungal plant pathogens. Compared to the FET3 protein from S. cerevisiae and Sclerotinia sclerotiorum, a closely related necrotrophic plant-pathogenic fungus, BcFET1 shares 47.5 and 87.2% amino acid sequence identity, respectively. According to the Conserved Domain Database (CDD) search (43), BcFET1 presents an MCO domain (COG2132) with three cupredoxin domains between amino acid residues 22 and 500 (Fig. 1b). Manual inspection of the alignment also allowed for the identification (in BcFET1) of a glutamic acid and a tyrosine residue equivalent to that of the FET3 protein from S. cerevisiae (185 and 354, equivalent to 189 and 357 in BcFET1; Fig. 1b). These two residues are critical for the ferroxidase activity of the protein (44). Collectively, and considering that the latter residues are absent from laccases, including BcLcc1 to BcLcc2, we conclude that bclcc13 Amino acid sequences were acquired from different filamentous fungal pathogens, as mentioned in the text. The analysis was performed using the maximum likelihood method on the Phylogeny.fr platform. The tree denotes MCO sequences as BcLcc1-12 and Bcin02g02780 (BcFET1), followed by their corresponding unique IDs. S. cerevisiae and S. sclerotiorum FET3 sequences (ScFET3, EDN96701) are also indicated (in boldface; GenBank protein accession number CAA89768). Accession numbers were also retrieved from protein FASTA files available at Ensembl Fungi. (b) Schematic representation of BcFET1 (top) and BcFTR1 (bottom). The L1 to L4 signatures of BcFET1 (615 aa; see the text) are denoted in green boxes, while MCO domains 1, 2, and 3 (PFAM PF00394, PF07731, and PF07732, respectively) are indicated in blue boxes. Predicted transmembrane domains (TMD) are also indicated. In the case of BcFTR1 (401 aa), seven transmembrane helices (TMH) were predicted and are indicated in orange boxes. Each protein is depicted to scale.  (39), bcfet1 is located in a genomic region close to its functional partner, here named bcftr1, which encodes an iron permease protein (gene identifier [ID] Bcin02g02790). As shown in Fig. S2, both bcfet1 and bcftr1 are 1,977 bp apart (between each translation start codon) and are divergent in transcriptional orientation. Computational analysis indicates the predicted BcFTR1 protein possesses 401 aa, shares 44% amino acid sequence identity with the FTR1 protein from S. cerevisiae, and presents the expected seven transmembrane helices (Fig. 1b, TMH; between aa residues 10 and 32, 53 and 75, 90 and 112, 153 and 175, 185 and 207, 214 and 236, and 295 and 317), as predicted by the TMHMM software (45). This allows the protein to channel ferroxidase (FET)-derived iron from the extracellular medium to the intracellular compartment. No paralog sequences were identified in the genome database.
Altogether, the provided in silico evidence strongly suggest the presence of an RIA system in B. cinerea that is encoded by gene IDs Bcin02g02780 and Bcin02g02790, composed of a single MCO gene encoding a BcFET1 ferroxidase and a functionally related gene encoding an iron permease, bcftr1.
Generation of bcfet1 deletion and complementation mutants. To assess the influence of BcFET1 on B. cinerea iron responses and virulence, mutants with disrupted bcfet1 activity were functionally characterized. For this purpose, we generated the Δbcfet1 mutant strain employing a homologous recombination strategy, which is illustrated in Fig. S3. Three independent Δbcfet1 deletion mutants with single integrations of the hph replacement cassette, as determined by quantitative PCR (qPCR) (see Table 2), were used. While all independent Δbcfet1 mutants exhibit equivalent phenotypes, an arbitrarily chosen mutant was subjected to genetic complementation to confirm the phenotypes associated with the deletion of bcfet1. Under the control of the endogenous wild-type bcfet1-bcftr1 divergent promoter region (Fig. S2), bcfet1 was targeted to the wild-type locus by homologous recombination. This yielded a hygromycin-sensitive and nourseothricin-resistant homokaryotic complementation mutant strain, termed Δbcfet1ϩbcfet1. As in the case of the Δbcfet1 mutant strain, the homologous recombination of the complemented mutant was analyzed by PCR (Fig. S4).
Deletion of bcfet1 affects regular conidium formation. As observed in Fig. 2, the Δbcfet1 strain displays a reduced and delayed conidiation pattern compared to that of the B05.10 wild-type strain ( Fig. 2a to d), which was determined as early as 3 and 4 days postinoculation (dpi) and also was supported by conidium quantification (Fig. S5). No significant differences in growth rate were observed after 3 days of growth, as depicted in Fig. S6. Interestingly, in addition to the reduced conidiation phenotype, at 4 dpi the Δbcfet1 strain started displaying primordia of sclerotia, which was observed as small white structures located near the inoculation site at the center of the petri dish (Fig. 2d). These structures were clearly observed 7 dpi ( Fig. 2f and h). After 7 days of cultivation in regular PDA medium, the B05.10 wild-type strain presents the expected conidiation pattern (Fig. 2e) covering the full cultivation area, while in the presence of plantsupplemented PDA media (PDAB), the B05.10 strain displays the well-known profuse conidiation pattern (Fig. 2g). In contrast, although the Δbcfet1 strain can produce conidia under both culture conditions, primordia of sclerotium structures are still observed. Notably, these structures were detected when employing the regular photoperiod commonly used to grow B. cinerea (see Materials and Methods), a culture condition that favors conidium, but not sclerotium formation, the latter being favored in the absence of light (13).
The deletion of bcfet1 leads to sclerotium formation and reduces whole-cell iron content. To investigate whether the white structures depicted in Fig. 2 lead to the development of regular dark-pigmented sclerotia, longer cultivation times were analyzed. As shown in Fig. 3, sclerotia were observed after 10 dpi for the Δbcfet1 mutant but not the B05.10 wild-type strain. These structures are required for overwintering, enabling the fungi to survive unfavorable conditions during extended periods of time when temperatures and/or nutrient availability are low (46). This observation is consistent with a mutant strain bearing a reduced/impaired acquisition of iron. Therefore, we tested whether the supplementation of the culture media with this micronutrient reduced sclerotium formation. Importantly, in the tested culture conditions, the wildtype strain does not produce sclerotia (Fig. 3a, top). As shown in Fig. 3a, the Δbcfet1 strain showed a reduced number of sclerotia in the iron-supplemented PDA medium, and these significant differences were dependent on iron concentration (Fig. 3b). As definitive proof that the absence of bcfet1 produces iron scarcity, we measured fungal iron content (see Materials and Methods). As shown in Fig. S7, after 3 dpi, the Δbcfet1 strain displayed a significantly reduced whole-cell iron content compared to that of the B05.10 wild-type strain. Finally, as shown in Fig. S8, the expression of bcfet1 in the Δbcfet1 genetic background reverted the sclerotium phenotype seen in the Δbcfet1 mutant strain, although the complemented mutant failed to display normal levels of conidiation (see Discussion).
The absence of bcfet1 leads to a hypervirulence phenotype that depends on the plant's iron status. To determine the relevance of BcFET1 and the RIA system during B. cinerea infection, we assayed virulence of the Δbcfet1 strain on French bean (Phaseolus vulgaris) and Arabidopsis thaliana Col-0 (Columbia 0) plants. These organisms represent highly and moderately Botrytis-susceptible hosts, respectively. As observed in Fig. 4, significantly increased necrotic lesions were found in both P. vulgaris and A. thaliana plants that were incubated for 3 days with the Δbcfet1 strain compared to the level for the B05.10 wild-type strain. In each case, plants were grown on a solid substrate mixture (details are in Materials and Methods). To properly control plant micronutrient availability, A. thaliana plants were grown hydroponically for 5 weeks and contained 50 M FeNa-EDTA as the only iron source, a strategy that has been previously reported (47). Consistent with the results obtained using French beans and A. thaliana grown on a solid substrate, hydroponically grown Arabidopsis plants that were grown under sufficient iron concentrations also displayed a significantly increased necrotic lesion when inoculated with the Δbcfet1 strain (Fig. 4d).
Taking advantage of the aforementioned hydroponic culture conditions, A. thaliana plants were subjected to an iron deprivation protocol (47) that used bathophenanthroline disulfonic acid (BPDS; see Materials and Methods), a highly specific iron chelator. Importantly, this method allows for normal Arabidopsis development after 4 weeks of cultivation; at this time, iron-deprived plants were obtained and used for virulence assays. As shown in Fig. 5, the iron deprivation procedure did not significantly affect the observed lesion area caused by the B05.10 wild-type strain. In contrast, when employing iron-starved plants, a significant reduction in the necrotic lesion caused by the Δbcfet1 strain was clearly observed, indicating that the hypervirulence phenotype determined for this particular mutant strain depends on the plant's iron availability.
The ⌬bcftr1 strain displays a wild-type level of infection. According to the Pathogen-Host Interaction (PHI) database (48), the majority of the B. cinerea mutant strains that have been characterized display reduced or unaltered virulence. Increased lesions are seldom observed, which makes the Δbcfet1 phenotype all the more interesting. Thus, to better analyze this very rare infection phenotype, the RIA system of B. cinerea was further characterized. To this end, two independent bcftr1 deletion mutants (Fig. S9) containing a single integration of the homologous hph recombination cassette (see Table 2) were obtained and characterized.
Similar to the Δbcfet1 strain, the bcftr1 mutant strain displays a reduced conidiation pattern that was clearly observed at the petri dish inoculation site (at the center of the plate). Nevertheless, no sclerotium formation was detected (Fig. 6a). Thus, this pheno- type can be considered less severe than that observed from the ⌬bcfet1 strain, although the ⌬bcftr1 mutant also shows reduced whole-cell iron content (Fig. S7). In contrast to the Δbcfet1 strain, the Δbcftr1 mutant displayed a wild-type level of infection in P. vulgaris plants (Fig. 6b). Taking advantage of the Arabidopsis iron deprivation procedure mentioned above, we also analyzed the virulence of the Δbcftr1 mutant strain using hydroponically grown plants. As observed in Fig. 4, a 2-fold increase in the necrotic lesion area was observed for the Δbcfet1 mutant (Fig. 6c) only in Arabidopsis plants obtained under sufficient iron concentrations, whereas the Δbcftr1 strain failed to do so. In contrast, the Δbcftr1 strain was found to display a wild-type level of infection in the case of iron-supplemented plants and resulted in significantly reduced lesions when infecting iron-deprived plants (Fig. 6c).
Considering that the infection phenotypes displayed by both RIA mutants of B. cinerea differ significantly, and because ROS is a hallmark of the plant defense responses, B. cinerea virulence (49), and iron homeostasis (50), we analyzed potential differences in the ROS levels caused by these two mutants during infection. To observe the oxidative burst (H 2 O 2 production) generated on the plant tissue infected by B. cinerea (51), DAB staining was performed. In addition, fungal structures were stained using cotton blue, which allows for fungal chitin staining (52). As observed in Fig. 7, a diminished level of DAB staining was found for all B. cinerea strains when infection assays were carried out on iron-starved plants (Fig. 7a, bottom). In agreement with this observation, in planta hyphal growth (determined with cotton blue staining) was also strongly reduced in iron-deprived plants. In contrast, and consistent with the quantification of the necrotic lesion area determined above, both B05.10 and Δbcftr1 strains developed a similar brown-stained area when infecting Arabidopsis plants obtained under iron sufficiency conditions (Fig. 7b). On the other hand, the Δbcfet1 strain consistently developed a larger area (2-fold increment; Fig. 7b) of DAB staining, but only when infecting iron-supplemented Arabidopsis plants.

DISCUSSION
Iron and the acquisition systems of this micronutrient have been described as modulators of infectious capacity. This includes different pathogens in general and phytopathogens in particular, including various fungi (36). However, iron capture systems in B. cinerea have not been studied in detail (39). The exceptions correspond to the results presented here and the identification of the major siderophores produced by this fungus, which were described several years ago, as well as their putative biosynthesis pathways, determined by computational predictions (29,30). For these reasons, at least from a fungal perspective, little is known about how this metal modulates the virulence of this remarkable necrotrophic phytopathogen. While the relevance of iron on plant defense responses has been investigated (see below), the role of the metal on B. cinerea virulence is just beginning to emerge.
Iron significantly affects plant defense responses to bacterial and fungal necrotrophic phytopathogens, including B. cinerea (21,47). The metal is an active player that participates in the defense response of A. thaliana, a plant that, when challenged with a pathogen, modifies iron uptake and mobilization (53)(54)(55). In this regard, at least two plant phytohormones, salicylic acid (SA) and jasmonic acid (JA), constitute major signaling pathways involved in plant defense and have a substantial impact on the  plant's iron homeostasis. Generally speaking, SA is involved in the signaling of defense mechanisms against biotrophic pathogens (56), while JA is considered the main phytohormone responsible for inducing the defense response to necrotrophs (57,58). The latter phytohormone, among the orchestrated plant defense responses, decreases iron capture (at the rhizosphere level) in Arabidopsis by downregulating the expression of FRO2 and IRT1, two genes that encode an iron uptake system. This strategy seeks to reduce the amount of iron that the pathogen can potentially acquire during infection (21,54). Consistent with these investigations and others (47), we found a reduced necrotic lesion when the B05.10 wild-type strain infects iron-starved plants, an observation that was even more pronounced (and statistically significant) in the case of both RIA knockout mutants of B. cinerea (Fig. 6). These results indicate the metal and the RIA system are necessary (from a fungal perspective) during plant infection. In agreement with these observations, a reduction in the leaf area that displays the oxidative burst during plant colonization was also determined but only in iron-deprived Arabidopsis. When the metal is not limited, however, it is reasonable to expect that, from the Δbcfet1 strain that is deficient in iron acquisition (as demonstrated in vitro), there will be reduced necrotic lesions compared to the wild-type strain or even similar necrotic lesions, if the absence of bcfet1 is compensated (e.g., siderophores), but not increased lesions, like those observed here.
Irrespective of the infection strategy employed by plant pathogens, both biotrophs and necrotrophs need iron to grow and survive during the infection. If the host withholds iron, virulence will decrease in both cases, although the plant needs iron to create an oxidative defense burst. In fact, SA increases iron uptake in Arabidopsis (21). However, necrotrophs like B. cinerea favor an oxidative environment, since it provides them with an advantage over biotrophs (49,59). In this regard, the mechanisms of iron acquisition employed by different pathogenic fungi seem to be closely related to their infection strategy, as previously noted (38). The uptake of iron mediated by siderophores has been described as fundamental for virulence in necrotrophic phytopathogens such as Alternaria brassicicola, Alternaria alternata, C. heterostrophus, F. graminearum, and other opportunistic mammalian pathogens that destroy the host cell, such as Aspergillus fumigatus (33,38,(60)(61)(62), and even entomopathogenic fungi (63). In contrast, biotrophic organisms that, unlike necrotrophs, require a living host cell to infect have been found to need the RIA system for iron acquisition during plant infection (see below), an observation that differs from the results reported here for B. cinerea.
Importantly, the complemented Δbcfet1ϩbcfet1 mutant clearly indicates that the expression of bcfet1 is required to inhibit sclerotium formation, structures that are no longer developed in the Δbcfet1 genetic background when the metal is added to the culture media. Nevertheless, the complemented mutant failed to restore regular conidiation, indicating that proper bcfet1 expression also is required for fungal morphogenesis and development. Considering that iron acquisition in fungal biological systems is delicately and tightly controlled at multiple levels, including the regulation of transcription (28,39), the strategy utilized to complement the wild-type copy back in the mutant background may have altered this. Indeed, based on the divergent transcriptional orientation of the two genes under analysis, it was not possible to employ the nat resistance cassette upstream of the 5= recombinational flank of bcfet1, in which case the divergent promoter region would have been interrupted. However, localizing the nat resistance cassette in the 3= region of the introduced copy of bcfet1 may have altered proper termination or mRNA accumulation. Nevertheless, the complementation strategy did allow us to confirm the observed hypervirulence effect associated with the absence of bcfet1 in the corresponding mutant.
The two most completely analyzed MCOs from B. cinerea are the laccases BcLcc1 and BcLcc2 (40). These types of enzymes can detoxify plant-derived antifungal and antimicrobial compounds such as phytoalexins (64), although bclcc2 loss-of-function mutants display wild-type levels of virulence (40), which sharply contrasts with the Δbcfet1 strain. The infection phenotype of this particular mutant strain also differs from those of other fungal pathogens in which the RIA system has been studied. Components of the RIA in different fungal pathogens have been described as crucial for observing their full pathogenic potential, which contrasts with the hypervirulence phenotype described here for the Δbcfet1 strain. This is the case for Candida albicans (65), in which RIA mutants display reduced virulence. In the case of the maize-specific biotrophic plant pathogen U. maydis, in the absence of the RIA system, it exhibited only a reduction in symptoms and plant lesions (37). However, the fungus still manages to cause disease, indicating that reductive iron assimilation is required to display full virulence. The same observation is valid for the biotrophic plant smut fungi M. violaceum (32). More recent investigations performed in the hemibiotrophic fungus Colletotrichum graminicola support the idea that the ferroxidase/permease system is required for virulence in biotrophic infections. In this fungus, fet3 is dispensable for the development of necrotic lesions on wounded plants, but the gene is required for the appressoria function, a key virulence determinant required for penetration of healthy plant tissue (66). This investigation and others (67) have provided evidence indicating that during the biotrophic phase of infection, RIA is used. When the fungus switches to the necrotrophic phase, siderophores are required for lesion development. Finally, recent investigations have provided additional perspectives for RIA. In Paracoccidioides species, whose genome does not encode an FTR1 iron permease, the authors suggested that the fungus utilizes a nonclassical FTR1-independent RIA system that requires Fe/Zn permeases, known as Zrts, that may account for iron uptake (68), adding an additional layer of complexity.
In S. cerevisiae, the FET3 and FTR1 proteins are not independent. They physically interact at the cytoplasmic level, forming a primary protein complex and subsequently comigrating to the plasma membrane, as demonstrated using fluorescence-tagged versions of each protein (69). In S. cerevisiae, it has been shown that the absence of the FTR1 protein causes a decrease in the migration of FET3 to the membrane with a fraction accumulating in cytoplasmic compartments. Similarly, in the absence of FET3, a partial migration of the iron permease to the membrane has been described (69). While this exemplifies the high level of functional interdependence necessary for metal uptake, it also provides genetic evidence that the components of the RIA system of B. cinerea are more autonomous than their yeast counterparts, since both RIA mutants display distinct (infection) phenotypes. As a multicopper ferroxidase that requires three Cu 2ϩ atoms (70), it has been reported that the yeast FET3 apoprotein is not adequately assembled and is unable to migrate to the membrane (69). The copper atom donation is carried out by the copper transport ATPase termed CCC2 (71), whose ortholog gene in B. cinerea has been designated bcccc2. This gene is required for copper incorporation, but importantly, the Δbcccc2 mutant is not found to be deficient in iron incorporation (72), which strongly suggests that the fungus possesses functionally alternative mechanisms to acquire iron. However, the precise biochemical pathways involved have not been experimentally validated yet (29,30).
According to the Pathogen-Host Interaction (PHI) database, and since the first successful genetic transformation in B. cinerea (73), about 150 loss-of-function mutants have been characterized and systematically organized. According to the PHI database, the vast majority (62.73%; 101 genes) of the available B. cinerea mutants lead to loss-of-pathogenicity and reduced-virulence phenotypes, with 34.16% of the mutants (55 genes) displaying unaffected pathogenicity and only 5 mutants (3.11%) having an increased virulence (hypervirulence) phenotype (48), a rare but increasingly prevalent infection phenotype among microbial pathogens (74). Any putative mechanistic connection among these five genes (75)(76)(77)(78) remains, until today, just speculation, although four out of these five mutants exhibited reduced or impaired conidium production. Interestingly, one of these genes (BcFKBP12) has been implicated in sulfur regulation (75), but its connection with iron metabolism and/or uptake is unknown.
It is logical to expect from an iron acquisition mutant a defect in the incorporation of this metal, as demonstrated here, and, concomitantly, reduced growth. If such a growth defect is observed in a pathogen, it is not easy to distinguish between reduced growth from decreased virulence, since the latter could be just the consequence of the former. Although to extrapolate in vitro growth (not altered in this study, at least under the conditions tested) to in planta growth behavior should be considered cautiously (e.g., virulence factors are expressed during the infection), it is fascinating to imagine a hypervirulence scenario caused by a growth-impaired mutant. In this regard, a reduction in the cotton blue staining was observed in iron-deprived plants for all three B. cinerea strains. In comparison, the two RIA mutant strains developed similar cotton blue staining when infecting replete iron plants (as seen in Fig. 7), reinforcing the interpretation of the enhanced virulence phenotype displayed solely by the lack of one of the RIA components. Thus, the key question is why does only the Δbcfet1 strain, and not the Δbcftr1 strain, display a hypervirulence phenotype, producing larger necrotic lesions?
Although RIA relies on and has been traditionally seen as an interdependent twocomponent system (the ferroxidase and the iron permease) (28), ferric reductases are needed to remove iron from siderophores (79), not only to provide Fe 2ϩ , especially under aerobic conditions, to the FET3 ferroxidase that produces Fe 3ϩ , which is subsequently incorporated by the FTR1 iron permease (Fig. 8a). In the absence of bcfet1, one or more iron reductases (the genome of B. cinerea encodes at least four of these proteins that are membrane bound) could favor the iron chemical equilibrium toward the reduced form of the metal, which may generate a stronger oxidative response due to the Fenton reaction (24). In the context of the plant oxidative burst associated with the defense response, this scenario may even be facilitated by the production of H 2 O 2 in a fungal infection that can completely or partially overcome this oxidative response (49, 59) to finally take advantage of ROS, facilitating the production of necrotic lesions (Fig. 8c). In contrast, in the absence of bcftr1 ROS is not favored, since the ferroxidase The reductive iron assimilation (RIA) system of B. cinerea is mainly composed of BcFET1 (brown membrane-anchor protein) and BcFTR1 (seven transmembrane helices protein). Environmental (chelated) Fe 3ϩ (orange dots) is reduced by iron reductases (FREx; blue). The genome of B. cinerea encodes at least four of these membrane-bound proteins, producing Fe 2ϩ that is subsequently oxidized by BcFET1 (Fe 3ϩ ; yellow dots). Finally, BcFTR1 incorporates iron into the intracellular compartment. A wild-type level of infection is depicted as a gray lesion over the plant's surface. (b) In the absence of BcFTR1, the ferroxidase BcFET1 produces Fe 3ϩ (yellow) that can no longer be incorporated as described for panel a due to the absence of the iron permease. A wild-type level of infection is observed. (c) In the context of the plant defensive oxidative burst and H 2 O 2 generation, the absence of BcFET1 favors the iron chemical equilibrium toward the reduced form of the metal, generating a stronger (Fenton) oxidative response and the hypervirulence phenotype of infection.
still produces the oxidized form of the metal that can no longer be incorporated due to the absence of the permease (Fig. 8b).
As demonstrated here, a loss-of-function mutant unexpectedly displays a significantly increased virulence phenotype, providing us with a unique opportunity to identify new virulence determinants in this genetic background. Genome-wide transcriptomics experiments are being carried out that deal with both defense and infection strategies, as well as the analysis of other iron acquisition components, such as membrane-bound iron reductases, to determine if this is the case.

MATERIALS AND METHODS
B. cinerea strain and culture conditions. B. cinerea Pers. Fr. (Botryotinia fuckeliana [de Bary] Whetzel) strain B05.10 was isolated initially in Germany from a Vitis vinifera vineyard (80). This strain was used as the recipient for genetic modification. Its genome sequence was initially published (4) and significantly improved as a final gapless high-quality genome assembly (81).
The B. cinerea B05.10 wild-type strain, as well as genetically modified strains, were cultivated in petri plates (100 mm in diameter) containing one of the following agar-containing media: potato dextrose agar (PDA; Becton, Dickinson) with and without 10% homogenized bean leaves (termed PDAB; cultivated in greenhouses) or Gamborg B5 (Duchefa Biochemie) supplemented with 2% glucose, employed during the genetic transformation procedure (see below). Depending on specific culture conditions, PDA plates were also supplemented with sterile FeCl 3 (in a range between 100 and 750 M). Since plants (see below) were routinely cultivated by employing a defined photoperiod, B. cinerea strains were also incubated in a 12:12-h light:dark regimen at 20°C using Percival incubators, as described previously (13).
Cloning of gene replacement and complementation cassettes. Two replacement cassettes (for Δbcfet1 and ⌬bcftr1) were assembled using yeast recombinational cloning (YRC), as described previously (82). Accordingly, the 5=-and 3=-noncoding regions of bcfet1 and bcftr1 were PCR amplified from the genomic DNA of B. cinerea B05.10 using the primer pairs indicated in Fig. S3 and S8. The hygromycin (hph) resistance cassette was amplified from the pCSN44 plasmid (obtained from the Fungal Genetics Stock Center [83]). Primers employed for these PCRs (indicated in the supplemental material) contained 30-bp 5=-overhang regions to allow homologous recombination by YRC. Fragments were cotransformed with the linearized pRS426 vector (84) into uracil-auxotrophic S. cerevisiae strain BY4741 for assembly (85). After YRC, each plasmid containing the desired genetic construct was recovered from individual yeast colonies by Escherichia coli (DH5␣) transformation. In the case of the bcfet1 gene complementation construct (see below), the open reading frame (ORF) of the mentioned gene was amplified from genomic DNA using the primer pair indicated in Fig. S4 (oligonucleotides pc05 and pc59), targeting the "wildtype" Δbcfet1 locus and employing the nourseothricin (nat) resistance cassette (obtained from the pNR1 vector [86]). The entire genetic construct was sequenced to confirm the absence of mutations. Thereafter, in each case, the replacement/complementation cassettes were PCR amplified using the external primers flanking each construct and the Kapa HiFi DNA polymerase (Kapa Biosystems). These PCR products were used for B. cinerea transformation.
Generation of B. cinerea deletion and complemented mutants. Protoplast generation and transformation of B. cinerea were performed as described previously (87) based on the original transformation protocol (73). In brief, B. cinerea conidia from a 1-week-old PDAB culture were incubated for 18 h at 20°C and 120 rpm in 100 ml malt extract medium (1.5%). After this period of cultivation, protoplasts from the young mycelia were generated using an enzymatic mixture containing 100 mg of lysing enzyme (Sigma) and 0.5 mg of yatalase (TaKaRa). The obtained protoplasts were mixed with 30 l of purified PCR products (mentioned above) in polyethylene glycol (PEG) solution (25% PEG 3350, 1 M CaCl 2 , 1 M Tris-HCl, pH 7.5). After 20 to 24 h of fungal regeneration on SH agar, they were overlaid with SH agar containing 70 g/ml hygromycin B (Invitrogen). Homokaryotic derivates were achieved after iterative steps of single-spore isolation on Gamborg B5-2% glucose supplemented with 70 g/ml hygromycin B and subsequent transfer of single colonies to new petri dishes. Following DNA extraction (88), transformants that have undergone homologous integration were confirmed by PCR using locusspecific primers (indicated in each corresponding figure in the supplemental material; mentioned above) in combination with hph-specific primers. The absence of wild-type alleles was confirmed by using "inner" primers designed to amplify the substituted genomic region, as depicted in Fig. S3, S4, and S8. For bcfet1 complementation of the Δbcfet1 mutant strain, protoplasts of the deletion mutant were transformed with the genetic construct depicted in Fig. S4 and overlaid with SH agar containing 140 g/ml nourseothricin (Goldbio). Targeted integration of the construct at the bcfet1 locus was detected using locus-specific primers, as well as nourseothricin (nat) resistance cassette-specific oligonucleotides.
Virulence assays employing Arabidopsis thaliana and Phaseolus vulgaris. Virulence assays were conducted as described previously (13). Essentially, conidia were suspended in Gamborg B5-2% glucose medium (B5) and adjusted to a final concentration of 2 ϫ 10 5 conidia/ml in B5 with 10 mM KH 2 PO 4 -K 2 HPO 4 , pH 6.4. Conidial suspensions (7.0 l) were used to inoculate leaves of Arabidopsis thaliana (Col-0) or primary leaves of French bean (Phaseolus vulgaris cv. Venus from the Instituto Nacional de Investigaciones Agropecuarias [INIA]-Chile). Approximately 5-week-old Arabidopsis plants were used. Arabidopsis plants were propagated in a 12:12-h light:dark regimen at 20°C using Percival incubators as described previously (13), while P. vulgaris plants were grown for 8 days with artificial LED light employing the mentioned photoperiod at 22°C inside a temperature-controlled growth chamber. For virulence assays, plants were incubated inside plastic boxes at 20°C under a humid environment within Percival incubators for 3 days. When indicated, the commercial substrate green power complete mix (TopCrop) was used, arranged in round plastic planters. This substrate is composed of 35% humus, 10% perlite, 10% vermiculite, 30% coconut fiber, and 15% blond peat. The developed lesions were recorded with a digital camera 72 h postinoculation (hpi) on both Arabidopsis and bean leaves. The size of each lesion was calculated by employing a semiautomated procedure using the ImageJ software (89) and an external calibration scale.
To assess the influence of iron on the infection caused by B. cinerea, A. thaliana plants were grown hydroponically to properly control iron availability. For this purpose, Arabidopsis plants were grown in a hydroponic medium for 5 weeks (50 M FeNa-EDTA) to allow normal development and then washed with 0.30 mM BPDS (bathophenanthroline disulfonic acid). The details of the mentioned strategy were previously described and were performed accordingly (47). In short, 5-day-old seedlings of A. thaliana were transplanted from petri dishes containing solid MS medium to an inert substrate (wool rock; GRO-SLAB) moistened with the following hydroponic medium: 0. 25 (47) and transferred back to a fresh hydroponic medium without iron. On the other hand, the control group was subjected to the same washing treatment but without BPDS, transferring these plants to fresh hydroponic medium containing 50 M FeNa-EDTA. Plants were incubated for 4 days under the conditions just described before performing virulence assays.
Real-time qPCR. PCR-verified mutants were further analyzed by qPCR to determine the number of hph cassette insertions. For this purpose, two pairs of primers were used ( Table 1) that amplify a defined and specific segment of the bcfrq1 gene of B. cinerea and hph. The B. cinerea bcfrq1 gene was chosen because it is a single-copy gene in the genome whose qPCR primers display very high qPCR amplification efficiency, as previously demonstrated (13) and validated here (Table 1). For both pairs of primers, each corresponding PCR amplicon was synthesized in vitro as a single-strand DNA molecule (Table 1; denoted amp97-98 and amp32-687). Employing known concentrations of these single-strand DNA templates (or amplicons) and 1:10 serial dilutions, the qPCR calibration curves were constructed as recommended (90), starting in each case from 500,000 molecules/l. Parallel qPCRs, employing genomic DNA derived from each analyzed mutant strain, were also performed. The qPCR crossing point (Cp) values obtained for the genomic DNA samples of each mutant isolate were interpolated into the mentioned calibration curves to determine the number of amplified DNA molecules, thereby defining the hph/bcfrq1 ratio (Table 2). For the qPCRs, the commercial SYBR FAST qPCR system (Kapa Biosystems) was used according to the manufacturer's directions. The qPCRs were carried out in the AriaMx real-time PCR system (Stratagene)  using the following amplification profile: initial denaturation at 95°C during 60 s, followed by 40 two-step qPCR of 5 s at 95°C and 15 s at 62°C. Determination of iron content in whole-cell fungal tissue. To determine the iron content in whole-cell fungal tissue, B05.10, Δbcfet1, and Δbcftr1 strains were grown on petri dishes containing PDA medium covered with cellophane paper to allow easy recovery of the fungal material. Tissue was collected in 2.0-ml Eppendorf tubes after 3 dpi. The fungal tissue was lyophilized until fully dried at Ϫ44°C and 9 Pa and subjected to weight determination. Completely dehydrated samples were digested and homogenized in 1.0 ml of 3% nitric acid during 16 h at 95°C by following the previously described procedure (91). The iron content present in the solution was subsequently determined by an atomic absorption spectrometer (Shimadzu AA-7000) employing an iron(III) (FeCl 3 ) calibration curve (Sigma).
DAB and cotton blue staining. H 2 O 2 levels in planta were determined by applying 3,3=diaminobenzidine (DAB) staining as described previously (51). Briefly, detached leaves were incubated in a 2 mM EDTA solution, pH 5.5, and subsequently incubated in a 5 mM DAB solution, pH 3.8, during 2 h with agitation. Leaves were destained in lactophenol. An additional cotton blue staining then was performed, allowing the detection of chitin, a polysaccharide that is present in the fungal cell wall (52,92). After that, images were acquired using an Olympus CX21 microscope attached to a digital camera.
Computational and statistical analyses. Phylogenetic reconstruction of MCO proteins encoded by the B. cinerea genome was performed by employing the one-click method of the Phylogeny.fr platform, applying default settings (42). For comparative purposes, several filamentous fungal RIA-related MCO ferroxidases were obtained from the fungal genomes of Aspergillus flavus, Claviceps purpurea, Colletotrichum orbiculare, Fusarium graminearum, Magnaporthe oryzae, Penicillium digitatum, Sclerotinia sclerotiorum, Trichoderma reesei, U. maydis, Verticillium dahliae, and Zymoseptoria tritici, available from FASTA files at Ensembl Fungi (93). Manual analysis and inspection of sequences were performed using ClustalW (94). GraphPad Prism software (version 8.3.0) was used for statistical analysis and data representation. Analysis of variance and Tukey's post test were performed (P Յ 0.05). A working model (depicted in Fig. 8) was constructed with the BioRender.com software.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.