Unconventionally Secreted Manganese Superoxide Dismutase VdSOD3 Is Required for the Virulence of Verticillium dahliae

: Plant pathogens generally employ superoxide dismutase (SOD) to detoxify host defense reactive oxygen species (ROS), and to scavenge ROS derived from their own metabolism. However, the roles of SODs in an important vascular pathogen, Verticillium dahliae , are unclear. Our previous study has shown that a putative signal-peptide-lacking manganese superoxide dismutase (VdSOD3) is present in the exoproteome of V. dahliae cultured in tissues of host cotton, suggesting that VdSOD3 may be exported out of the fungal cells and contribute to the SOD activity extracellularly. Here, we conﬁrm that the N -terminal of VdSOD3 is not a functional signal peptide by yeast signal trap assay. Despite lacking the signal peptide, the extracellular distribution of VdSOD3 was observed in planta by confocal microscopy during infection. Loss-of-function of VdSOD3 decreased extracellular and intracellular SOD activities of V. dahliae by 58.2% and 17.4%, respectively. Deletion mutant of VdSOD3 had normal growth and conidiation but showed signiﬁcantly reduced virulence to susceptible hosts of cotton and Nicotiana benthamiana . Our data show that signal-peptide-lacking VdSOD3 is a dual function superoxide dismutase, localizing and functioning intracellularly and extracellularly. Whereas nonessential for viability, VdSOD3 plays a vital role in the virulence of V. dahliae .


Introduction
In the course of infection, besides being challenged by reactive oxygen species (ROS) produced intracellularly as a natural byproduct of pathogen respiration, plant pathogens are additionally confronted with extracellular ROS generated by the host innate immune response [1]. These ROS have the oxidation capacity to damage virtually all organic and many inorganic components of the pathogen, leading to cellular dysfunction and eventually pathogen death [2]. Superoxide dismutases (SODs), the first-line enzymes against superoxide damage [3], are highly conserved metalloenzymes that use a redox-active cofactored metal to catalyze the conversion of superoxide anion to oxygen and hydrogen peroxide; the latter is further removed by catalase and peroxidase enzymes [4].
Manganese superoxide dismutase (Mn-SOD) is a SOD family member, which harbors Mn 2+ as the metal cofactor [5]. Significant progress has been achieved in the functional analysis of Mn-SODs of specific pathogens. Mn-SODs contribute to pathogens' virulence such as Cryptococcus neoformans, Metarhizium robertsii, Candida glabrata and Beauveria bassiana [6]. However, other reports demonstrate a nonessential role of Mn-SOD in the virulence of pathogens, such as Aspergillus fumigatus and Colletotrichum graminicola [6]. These studies demonstrate that Mn-SODs differentially contribute to virulence on plant hosts.
The currently available Mn-SODs are generally considered intracellular enzymes lacking N-terminal signal peptides [7]. However, clear evidence shows that bacterial Mn-SOD (termed SodA) is localized in the extracellular spaces. SodA has been reported to be exported to the periplasm or the culture supernatants of various bacteria [8][9][10]. In addition, the Mn-SOD activity can be detected from the supernatant of bacterial cultures [11]. Thus, Mn-SOD has been thought to be an unconventionally secreted protein in bacteria for a long time [12]. Fungi contain a Mn-SOD with a predicted mitochondrial targeting signal, typically called SOD2. In addition, a certain class of fungi also harbor a cytosolic Mn-SOD without this targeting signal, which is termed SOD3 [5]. Recently, increasing evidence suggests that fungal cytosolic SOD3 may also be exported extracellularly. For instance, SOD3 of certain fungi is recognized specifically by the host and is considered as a fungal allergen, most of which are extracellular proteins [13,14]. Moreover, proteomics analysis demonstrates that SOD3 is present in many fungal secretomes, such as Aspergillus fumigatus and Fusarium graminearum [15]. However, there is no concrete experimental evidence for the unconventional secretion of fungal SOD3.
Verticillium dahliae Kleb. is a destructive phytopathogenic fungus that attacks a wide range of hosts and targets the xylem tissue, causing Verticillium wilt disease on many economically important crops, such as cotton, potato and lettuce [16]. During the battle between V. dahliae and hosts, ROS are accumulated at the infected site, suggesting ROS are also involved in the fungus-plant interaction [17]. Recently, to identify new factors facilitating V. dahliae infection, we examined the exoproteome of V. dahliae induced in the medium containing tissues of susceptible host cotton by iTRAQ. From this, we identified a predicted signal-peptide-lacking Mn-SOD (VdSOD3) with seven unique matching peptides and 41.95% coverage of the full length of the VdSOD3 protein [18], suggesting that VdSOD3 is probably secreted out and functions extracellularly. However, the function of VdSOD3 during infection in host plants is yet to be understood.
Thus, the main objectives of the current study were: (1) to investigate whether VdSOD3 is a secretory protein; (2) to determine the SOD activity of VdSOD3; (3) to determine the role of VdSOD3 in pathogenicity.

Gene Cloning and Bioinformatics Analysis
The coding sequence of VdSOD3 (VEDA_05933) was amplified from cDNA samples of Vd991 with primers SOD3-F/R based on V. dahliae genome database [19]. All primers used for vector construction are listed in Table S1. Multiple sequence alignment was performed by Clustal X2. The N-terminal peptide of VdSOD3 was predicted using SignalP 5.0 [20]. Unconventionally secreted protein prediction was performed using OutCyte 1.0 [21].

Yeast Signal Sequence Trap System
The secretion function of the N-terminal signal peptide of VdSOD3 was tested by the yeast invertase secretion assay as previously described [22]. The coding region of the first 35 amino acids (aas) in the N-terminal peptide of VdSOD3 was amplified with primers N35-F/R and fused into the pSUC2 vector. The reconstruct plasmid was transformed into the yeast strain YTK12, and the positive clones were incubated on YPRAA medium (1% yeast extract, 2% peptone, 2% raffinose and 2 µg/mL antimycin A) to test the secretion ability of N-terminal peptide of VdSOD3. The known secretion function of the N-terminal peptide from Avr1b was set as the control.

Fungal Transformation
Targeted gene deletion plasmid was generated based on a previously described method with modifications [19]. Briefly, a 1242 bp upstream sequence, a 1178 bp downstream sequence of VdSOD3 and the hygromycin phosphotransferase gene (Hpt) cassette were fused to one DNA fragment via fusion PCR. Subsequently, the fusion product was then cloned into a Hind III/Xba I-linearized pGKO2-Gateway vector to generate the targeted gene deletion plasmid using a homologous recombination reaction. The complementation DNA fragment, containing the VdSOD3 wild-type gene with its native promoters and terminator regions, was fused into the binary vector pCOM with geneticin resistance to generate the complementation plasmid. A. tumefaciens-mediated fungal transformation was performed as described previously [23]. Gene knockout mutants were selected on PDA medium supplemented with hygromycin (50 µg/mL), while the complemented strains were selected in the presence of geneticin (50 µg/mL). Single spore isolations were performed for all transformants followed by PCR verification.
For expression of the VdSOD3-green fluorescent protein (GFP) fusion protein, the VdSOD3 ORF was fused into the Kpn I sites of pCOM-GFP (integrating the GFP ORF into pCOM) using the recombination method. The recombined plasmid was transformed into Vd991 to create WT::VdSOD3-GFP.

Confocal Microscopy Analysis
To examine the subcellular localization of VdSOD3, the coding sequence of VdSOD3 was recombined with pBin::GFP to construct pBin::VdSOD3-GFP. The plasmid was transformed into A. tumefaciens GV3101, and agroinfiltrated into 4-week-old N. benthamiana leaves. To observe green fluorescence, the infiltrated tobacco leaves were harvested 2 days post-agro-infiltration and then imaged under a laser scanning confocal microscope (LSM T-PMT) with excitation and emission wavelengths of 488 and 510 nm, respectively. To observe the localization of VdSOD3 in host-pathogen interactions, the inner layer of onion epidermal cells imitated the host plants to inoculate with conidial suspension (1 × 10 7 conidia/mL −1 ) of strain WT::VdSOD3-GFP. After incubating on a water agar (1%) plate at room temperature for 2 days, laser scanning confocal microscopy was used to view green fluorescence in onion epidermal cells.

VdSOD3 Expression Analysis
The V. dahliae Vd991 strain was cultured in liquid CM plus 1 mM MnCl 2 from 3 to 9 days. Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Potential genomic DNA was digested with DNase I (TaKaRa Bio, Shiga, Japan). First-strand cDNA was synthesized using a Super-Script III First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA, USA) with an oligo(dT)18 primer. The expression levels of VdSOD3 were measured by qRT-PCR using FastFire qPCR premix (SYBR Green, TianGen, Beijing, China). The real-time PCR program consisted of an initial denaturation step at 94 • C for 10 min, followed by 40 cycles of 94 • C for 15 s and 60 • C with the primer qSOD3-F/R for 1 min. The expression level of VdSOD3 was normalized to that of V. dahliae elongation factor 1-α (EF-1α). Each reaction was carried out in triplicate.

Measurement of SOD Enzyme Activity
The wild-type strain, ∆SOD3, and complemented strains were cultured in liquid CM plus 1 mM MnCl 2 for 7 days. The supernatant and fungal tissue were separated by centrifugation and ultrafiltration membrane filtration. The total SOD activity of each sample was measured following the nitro blue tetrazolium (NBT) reduction method [24]. The SOD unit was defined as the amount of crude enzyme required to inhibit the reduction in NBT by 50% and was expressed as units per mg protein (U/mg protein).

Fungal Morphology and Pathogenicity
For colony morphology and radial growth rate assay, a 2 µL conidial suspension with a concentration of 2 × 10 6 conidia/mL −1 of the wild-type strain Vd991, VdSOD3 deletion mutants and complemented strains was placed in the center of a PDA plate and CM plates containing different carbon sources (sucrose, starch, pectin and sodium carboxymethyl cellulose, 1% w/v), respectively. After incubating at 25 • C for 9 days, fungal colonies were measured and photographed. To evaluate conidial production, agar plugs were collected from the edges of 5-and 7-day-old fungal colonies using a 5 mm diameter cork borer. After shaking in 1 mL of sterile water, the number of conidia was quantified using a hemocytometer.
Pathogenicity assays were carried on N. benthamiana and cotton seedlings using a rootdip method as previously described [19]. Briefly, N. benthamiana and cotton seedlings were gently uprooted, immersed in 20 mL of 5 × 10 6 conidia/mL −1 suspension from V. dahliae strain for 5 min and then replanted into new pots with three replicates. In each replicate, 12 N. benthamiana plants or six pots of cotton seedlings (five plants for each pot) were used for each transformant. After maintaining at 28 • C (14 h/10 h, day/night cycle) for three weeks, disease phenotypes and vascular discoloration were photographed. After extraction of the genomic DNA of shoots of the infected plants, the fungal biomass was quantified by qPCR with V. dahliae EF-1α for quantifying fungal colonization by standardization with endogenous plant control of cotton 18S gene or N. benthamiana EF-1α.

Bioinformatics Analysis of VdSOD3
The V. dahliae genome possesses two divergent class of Mn-SODs: VdSOD2 with a predicted mitochondria-targeting signal (Gene-ID in Vd991 genome: VEDA_08263, Gene-ID in VdLs.17 genome: XP_009658205.1) and cysotolic VdSOD3 without this targeting signal (Gene-ID in Vd991 genome: VEDA_05933, Gene-ID in VdLs.17 genome: EGY16066.1) [16,19]. Proteomic analysis of the V. dahliae Vd991 extracellular proteins mimicking host-pathogen interactions identified VdSOD3 in the exoproteome of V. dahliae [18]. The VdSOD3 ORF was predicted to encode a polypeptide of 205 aas with a predicted molecular weight of 22.69 kDa and an isoelectric point of 6.60. Sequence alignment of VdSOD3 with other plant fungal pathogens' SOD3 displayed high homology, with 65.85% identity to Fusarium graminearum and 70.24% to Magnaporthe oryzae ( Figure 1A), respectively. VdSOD3 has four predicted metal binding sites for manganese (H27, H71, D156 and H160), a conserved DMWEHAYY region (156-163 aa) and Mn-SOD specific residues (G66, G67, F74, Q143 and D144) ( Figure 1A). Sequence analysis with SignalP 5.0 predicted that the N-terminus of VdSOD3 would not display signal peptide characteristics with high probability (0.9988) ( Figure 1B). However, VdSOD3 was predicted to be a potential unconventionally secreted protein by OutCyte 1.0 with a score of 0.6323, which is above the default threshold of 0.50 ( Figure 1C). Together, these results indicated that VdSOD3 is likely a signal-peptide-lacking Mn-SOD exported into the extracellular space unconventionally.

Signal-Peptide-Lacking VdSOD3 Can Be Secreted
Subsequently, the secretory ability of the N-terminus of VdSOD3 was determined by the yeast signal trap assay to confirm the prediction of SignalP 5.0. Yeast needs to secrete invertase to grow on the YPRAA medium, which contains raffinose as the sole carbon source. The 35 aas of the VdSOD3 N-terminal peptide (VdSOD3 N35 ), where the signal peptide is generally found, were fused to the vector pSUC2, which carries a signal-peptidelacking invertase gene. The recombined plasmid was then transformed into the invertase negative yeast strain YTK12 to create YTK12::VdSOD3 N35 . In the CMD-W medium with sucrose as the carbon source, YTK12::VdSOD3 N35 grew well. However, unlike the known signal peptide from Avr1b, VdSOD3 N35 cannot mediate invertase secretion to provide the viability of YTK12::VdSOD3 N35 in the YPRAA medium ( Figure 2A). Thus, these results indicated that VdSOD3 does not contain a functional signal peptide, as predicted by Sig-nalP 5.0.

Signal-Peptide-Lacking VdSOD3 Can Be Secreted
Subsequently, the secretory ability of the N-terminus of VdSOD3 was determined by the yeast signal trap assay to confirm the prediction of SignalP 5.0. Yeast needs to secrete invertase to grow on the YPRAA medium, which contains raffinose as the sole carbon source. The 35 aas of the VdSOD3 N-terminal peptide (VdSOD3 N35 ), where the signal peptide is generally found, were fused to the vector pSUC2, which carries a signal-peptidelacking invertase gene. The recombined plasmid was then transformed into the invertase negative yeast strain YTK12 to create YTK12::VdSOD3 N35 . In the CMD-W medium with sucrose as the carbon source, YTK12::VdSOD3 N35 grew well. However, unlike the known signal peptide from Avr1b, VdSOD3 N35 cannot mediate invertase secretion to provide the viability of YTK12::VdSOD3 N35 in the YPRAA medium ( Figure 2A). Thus, these results indicated that VdSOD3 does not contain a functional signal peptide, as predicted by SignalP 5.0. WT::GFP ( Figure 2C), indicating the cytosolic localization of VdSOD3. However, only the invaded onion epidermal cells inoculated with WT::VdSOD3-GFP showed strong green fluorescence around onion epidermal cells, while no fluorescence was observed in the onion tissues infected by WT::GFP ( Figure 2C), demonstrating that VdSOD3 is also a secreted protein, which translocates extracellularly during infection. Therefore, our results support V. dahliae VdSOD3 can be secreted extracellularly, although it lacks a functional signal peptide.

VdSOD3 Expression Pattern Analysis
Since the expression of SOD3 of Candida albicans was strongly stimulated upon entering the stationary phase in continuous cultures [25], the gene expression pattern of VdSOD3 was detected in time course (3-9 days) in continuous cultures. Similar to the To investigate whether VdSOD3 is secreted, the coding region of VdSOD3 was fused with the green fluorescent protein (GFP) and transiently expressed in N. benthamiana leaves by agroinfiltration. The results showed that, in addition to the cytoplasm, VdSOD3-GFP also aggregated at the periphery of N. benthamiana cells ( Figure 2B), indicating that Vd-SOD3 can be translocated extracellularly. Further, the onion epidermal cells, an advantage material to capture the GFP signal, imitated the host plant to determine the secretion characteristic of VdSOD3 when inoculated with V. dahliae. We constructed WT::VdSOD3-GFP, which expresses the VdSOD3-GFP fusion protein under the TrpC promoter in the wild-type strain background. WT::GFP, a V. dahliae wild-type strain expressing free cytosolic GFP, was used as a negative control. Conidial spores of WT::VdSOD3-GFP and WT::GFP were inoculated on onion epidermal cells for 2 days. As expected, the green fluorescence was observed in the fungal hyphae for both WT::VdSOD3-GFP and WT::GFP ( Figure 2C), indicating the cytosolic localization of VdSOD3. However, only the invaded onion epidermal cells inoculated with WT::VdSOD3-GFP showed strong green fluorescence around onion epidermal cells, while no fluorescence was observed in the onion tissues infected by WT::GFP ( Figure 2C), demonstrating that VdSOD3 is also a secreted protein, which translocates extracellularly during infection. Therefore, our results support V. dahliae VdSOD3 can be secreted extracellularly, although it lacks a functional signal peptide.

VdSOD3 Expression Pattern Analysis
Since the expression of SOD3 of Candida albicans was strongly stimulated upon entering the stationary phase in continuous cultures [25], the gene expression pattern of VdSOD3 was detected in time course (3-9 days) in continuous cultures. Similar to the SOD3 of Agronomy 2021, 11, 13 7 of 12 C. albicans, expression of VdSOD3 was also induced as the cells entered the stationary phase, reaching the maximum level at 7 days after culturing (Figure 3). SOD3 of C. albicans, expression of VdSOD3 was also induced as the cells entered the stationary phase, reaching the maximum level at 7 days after culturing (Figure 3).

VdSOD3 Contributes to Both Intra-and Extracellular SOD Activity of V. Dahliae
To further explore the function of VdSOD3, gene knockout mutant of VdSOD3 was constructed by replacing the coding sequence of the wild-type strain with hygromycin phosphotransferase gene (Hpt) cassette via homologous recombination. Five transformants were verified by PCR, of which two deletion mutants (ΔSOD3-1 and SOD3-2) were selected randomly for further research ( Figure 4A). Two randomly complemented strains (EC-1 and EC-2) were selected by reintroducing VdSOD3 with its native promoter and the neomycin resistance gene to ΔSOD3-1 ( Figure 4B).
Due to VdSOD3 being located both intra-and extracellularly in V. dahlia, as detected by a confocal microscope, the total SOD activity was tested for both the fungal hyphal and culture filtrates of the wild-type strain and ΔSOD3-1/2 7 days after culturing (strongly expression stage) in liquid CM plus Mn 2+ (for inducing Mn-SOD activity). As expected, we found that the total SOD activity in the hyphal extract decreased significantly by 58.23% in VdSOD3 deletion mutants compared to that of the wild-type strain, indicating that VdSOD3 contributed to the majority of fungal tissue SOD activity of V. dahliae under the tested conditions ( Figure 4C). Meanwhile, the SOD activity in fungal culture filtrates also decreased by 17.37% as a consequence of VdSOD3 deletion ( Figure 4C), suggesting that VdSOD3 also contributed to a portion of the SOD activity of the culture filtrates, probably due to its secretion to the extracellular. Complemented strains restored culture filtrates and fungal tissue SOD activities, which were comparable to those of the wild-type strain ( Figure 4C).

VdSOD3 Contributes to Both Intra-and Extracellular SOD Activity of V. Dahliae
To further explore the function of VdSOD3, gene knockout mutant of VdSOD3 was constructed by replacing the coding sequence of the wild-type strain with hygromycin phosphotransferase gene (Hpt) cassette via homologous recombination. Five transformants were verified by PCR, of which two deletion mutants (∆SOD3-1 and SOD3-2) were selected randomly for further research ( Figure 4A). Two randomly complemented strains (EC-1 and EC-2) were selected by reintroducing VdSOD3 with its native promoter and the neomycin resistance gene to ∆SOD3-1 ( Figure 4B).
Due to VdSOD3 being located both intra-and extracellularly in V. dahlia, as detected by a confocal microscope, the total SOD activity was tested for both the fungal hyphal and culture filtrates of the wild-type strain and ∆SOD3-1/2 7 days after culturing (strongly expression stage) in liquid CM plus Mn 2+ (for inducing Mn-SOD activity). As expected, we found that the total SOD activity in the hyphal extract decreased significantly by 58.23% in VdSOD3 deletion mutants compared to that of the wild-type strain, indicating that VdSOD3 contributed to the majority of fungal tissue SOD activity of V. dahliae under the tested conditions ( Figure 4C). Meanwhile, the SOD activity in fungal culture filtrates also decreased by 17.37% as a consequence of VdSOD3 deletion ( Figure 4C), suggesting that VdSOD3 also contributed to a portion of the SOD activity of the culture filtrates, probably due to its secretion to the extracellular. Complemented strains restored culture filtrates and fungal tissue SOD activities, which were comparable to those of the wild-type strain ( Figure 4C).  . Asterisks ** and * represent significant differences p < 0.01 and p < 0.05 by Student's t-test, respectively.

Fungal Morphology and Pathogenicity
The colony morphology and growth rate of VdSOD3 deletion mutants were comparable to those of the wild-type strain on PDA plates. Moreover, the mutants' growth rates were also identical with those of the wild-type strain on medium with sucrose, starch, pectin and cellulose as sole carbon sources ( Figure 5A). Additionally, the number of conidia produced by the two VdSOD3 deletion strains was similar to that of the wild-type strain ( Figure 5B). Thus, these results suggested that VdSOD3 is not required for fungal viability. . Asterisks ** and * represent significant differences p < 0.01 and p < 0.05 by Student's t-test, respectively.

Fungal Morphology and Pathogenicity
The colony morphology and growth rate of VdSOD3 deletion mutants were comparable to those of the wild-type strain on PDA plates. Moreover, the mutants' growth rates were also identical with those of the wild-type strain on medium with sucrose, starch, pectin and cellulose as sole carbon sources ( Figure 5A). Additionally, the number of conidia produced by the two VdSOD3 deletion strains was similar to that of the wild-type strain ( Figure 5B). Thus, these results suggested that VdSOD3 is not required for fungal viability. Further, to assess the possible role of VdSOD3 to virulence in V. dahliae, the susceptible cotton Gossypium hirsutum cv. Junmian No.1 was inoculated with wild-type strain Vd991, VdSOD3 deletion mutants and complemented strains, respectively, using a root-dip method. As expected, cotton inoculated with the wild-type strain showed leaf wilting, necrosis and vascular discoloration phenomenon, which are typical Verticillium wilt symptoms. In contrast, cotton inoculated with two VdSOD3 deletion strains displayed significantly alleviated symptoms of the disease ( Figure 6A), along with reduced fungal biomass in planta relative to cotton inoculated with the wild-type strain ( Figure 6C). Similarly, ∆SOD3inoculated N. benthamiana plants also displayed attenuated Verticillium wilt symptoms ( Figure 6B) and reduced fungal biomass compared with plants inoculated with the wildtype strain ( Figure 6D). Complemented strains showed that the disease symptoms and fungal biomass were comparable to those of the wild-type strain ( Figure 6A-D). Taken together, these results suggested that the unconventionally secreted protein VdSOD3 was required for full virulence of V. dahliae on the host plant of cotton and tobacco. Further, to assess the possible role of VdSOD3 to virulence in V. dahliae, the susceptible cotton Gossypium hirsutum cv. Junmian No.1 was inoculated with wild-type strain Vd991, VdSOD3 deletion mutants and complemented strains, respectively, using a rootdip method. As expected, cotton inoculated with the wild-type strain showed leaf wilting, necrosis and vascular discoloration phenomenon, which are typical Verticillium wilt symptoms. In contrast, cotton inoculated with two VdSOD3 deletion strains displayed significantly alleviated symptoms of the disease ( Figure 6A), along with reduced fungal biomass in planta relative to cotton inoculated with the wild-type strain ( Figure 6C). Similarly, ΔSOD3-inoculated N. benthamiana plants also displayed attenuated Verticillium wilt symptoms ( Figure 6B) and reduced fungal biomass compared with plants inoculated with the wild-type strain ( Figure 6D). Complemented strains showed that the disease symptoms and fungal biomass were comparable to those of the wild-type strain ( Figure 6A-D). Taken together, these results suggested that the unconventionally secreted protein VdSOD3 was required for full virulence of V. dahliae on the host plant of cotton and tobacco.

Discussion
Despite extensive studies on the secretory characteristics of signal-peptide-lacking Mn-SOD in bacteria [12], few studies have referred to the extracellular distribution of Mn-

Discussion
Despite extensive studies on the secretory characteristics of signal-peptide-lacking Mn-SOD in bacteria [12], few studies have referred to the extracellular distribution of Mn-SOD in fungi, except for the extracellular fungal Mn-SODs detected by proteomics analysis [15]. In addition, Mn-SOD enzymatic activity has only been identified from cell-free extracts of the filamentous fungus Penicillium chrysogenum [26]. In this study, bioinformatics analysis suggested that cytosolic VdSOD3 in V. dahliae could be secreted unconventionally (Figure 1). Although the yeast signal trap assay proved that VdSOD3 has no signal peptide, its translocation to host apoplastic spaces in vivo was observed by confocal microscopy (Figure 2), providing conclusive proof that VdSOD3 is an unconventionally secreted protein. Not surprisingly, VdSOD3 contributes to the SOD activity of the culture filtrate of V. dahliae (Figure 4). To our knowledge, this is the first study to identify both the secretory characteristics and extracellular catalytic function of Mn-SOD simultaneously in a fungal phytopathogen and offer a possible explanation of how Mn-SOD is involved in detoxifying extracellular ROS produced by the host.
Unconventionally secreted proteins involved in the virulence of fungal phytopathogens is an emerging theme in plant-microbe interactions. Secretion of plant cytoplasmic effectors in Magnaporthe oryzae is independent of the Golgi complex [27]. Signal-peptide-lacking sterol carrier protein 2 (Scp2) of Ustilago maydis was detected in the apoplastic fluid of infected maize, which plays a role in the inhibition of competitors in the apoplast [28]. Our previous work identified 99 signal-peptide-lacking proteins in V. dahliae secretome, implying that many unconventional secreted proteins may play essential roles during host-pathogen interactions [18]. However, only one signal-peptide-lacking protein, VdIsc1, a plant cytoplasmic effector regulating the hydrolysis of salicylate's precursor, has been extensively studied [29]. In this study, we identified another signal-peptide-lacking protein with plant apoplastic localization, furthering the understanding of how unconventionally secreted proteins contribute to the virulence of V. dahliae. We speculate that VdSOD3 following an unconventional secretion pathway may be a pathogen strategy to maintain Mn-SOD function both inside and outside of the V. dahliae cells.
We found that the expression of VdSOD3 was induced as the cells entered the stationary phase, similar to the SOD3 of C. albicans, which was also stimulated upon approaching the stationary phase in continuous cultures [25]. Recently, Culotta lab found that intracellular copper decreases during stationary phase of C. albicans and as part of a copper starvation response, C. albicans represses Cu/Zn SOD1, which requires Cu 2+ for SOD activity and switches to the noncopper alternative SOD3 [30]. Interestingly, the V. dahliae genome also harbors a putative Cu/Zn VdSOD1 (Gene-ID in Vd991 genome: VEDA_03436, Gene-ID in VdLs.17 genome: VDAG_02630) [16,19] and this principle may also apply to V. dahliae. VdSOD3 may substitutes for VdSOD1 when copper is low to ensure constant SOD activity of V. dahliae during different stages of infection.
Mn-SODs are important virulence factors in nearly all phytopathogens, with Mn-SOD of C. graminicola being the exception [6]. In this study, deletion of VdSOD3 did not affect the viability of V. dahliae, but alleviated the disease symptoms and reduced V. dahliae biomass in both cotton and N. benthamiana, suggesting that VdSOD3 was required for the virulence of V. dahliae ( Figure 6). Two possible reasons may explain this phenomenon. First, the significantly impaired intracellular SOD activity in ∆VdSOD3 may lead to fungal inability to manage its metabolic superoxide to promote pathogen survival within the host. Second, the extracellular distribution of VdSOD3 may function in scavenging extracellular host-derived superoxide, and ∆VdSOD3 may have difficulty in surviving the ROS storm generated by the host.

Conclusions
In addition to being localized inside the cell, signal-peptide-lacking manganese superoxide dismutase VdSOD3 can also secreted to extracellular and possesses the SOD activity inside and outside of the cells for host-V. dahliae interactions. While being nonessential for growth or conidiation, VdSOD3 contributes to the virulence of V. dahliae during infection in host plants.