Potential of Antifungal Proteins (AFPs) to Control Penicillium Postharvest Fruit Decay

Penicillium phytopathogenic species provoke severe postharvest disease and economic losses. Penicillium expansum is the main pome fruit phytopathogen while Penicillium digitatum and Penicillium italicum cause citrus green and blue mold, respectively. Control strategies rely on the use of synthetic fungicides, but the appearance of resistant strains and safety concerns have led to the search for new antifungals. Here, the potential application of different antifungal proteins (AFPs) including the three Penicillium chrysogenum proteins (PAF, PAFB and PAFC), as well as the Neosartorya fischeri NFAP2 protein to control Penicillium decay, has been evaluated. PAFB was the most potent AFP against P. digitatum, P. italicum and P. expansum, PAFC and NFAP2 showed moderate antifungal activity, whereas PAF was the least active protein. In fruit protection assays, PAFB provoked a reduction of the incidence of infections caused by P. digitatum and P. italicum in oranges and by P. expansum in apples. A combination of AFPs did not result in an increase in the efficacy of disease control. In conclusion, this study expands the antifungal inhibition spectrum of the AFPs evaluated, and demonstrates that AFPs act in a species-specific manner. PAFB is a promising alternative compound to control Penicillium postharvest fruit decay.


Introduction
Postharvest decay caused by phytopathogenic fungi provokes major economic losses for the worldwide industry of fresh horticultural products [1]. Furthermore, some fungal phytopathogens are responsible for the contamination of fruits and derivative products through the production of mycotoxins that are detrimental to human health [2]. Species from the Penicillium genus cause severe postharvest fruit diseases even when postharvest technologies are applied. Penicillium digitatum, the cause of citrus green mold, and Penicillium italicum, the cause of citrus blue mold, are the main responsible agents of postharvest citrus losses worldwide, since more than 90% of citrus rots are produced by these two species [3]. P. digitatum commonly causes larger losses during commercialization due to its predominance at ambient temperatures, while P. italicum decay is higher in cold-stored

Antifungal Assays
Susceptibility tests were carried out in 96-well, flat-bottom microtiter plates (Thermo Scientific, Waltham, MA, USA) as described before [15]. Briefly, 100 µL of conidia (1 × 10 4 conidia/mL) in 10% (v/v) PDB were mixed with 100 µL of AFP prepared in serial twofold dilutions (in 10% (v/v) PDB) to reach final concentrations of 0-64 µM. Plates were statically incubated for 72 h at 25 • C. Growth was determined every 24 h by measuring the optical density (OD) of the fungal cultures at 620 nm using FLUOstar Omega plate spectrophotometer (BMG Labtech, Ortenberg, Germany), and the OD 620 mean and standard deviation (SD) of three replicates were calculated. Minimum inhibitory concentration (MIC) was defined as the protein concentration that completely inhibited fungal growth ≥90% in all the independent experiments performed (n = 2-3).

Protection Assays against Fungal Infections Caused by Penicillium spp. in Fruits
Assays were conducted either with freshly harvested oranges (Citrus sinensis L. Osbeck cv Navel and Lanelate) or with apples (Malus domestica cv Golden Delicious) obtained from a local grocery. Fruits were surface sterilized by incubation for 5 min in a 5% commercial bleach solution and subsequent washing in distilled water three times, and air-dried. For protection assays, three replicates of five fruits were inoculated at four wounds around the equator with 5 µL of conidial suspensions (10 4 conidia/mL for P. digitatum and P. expansum, and 2.5 × 10 4 for P. italicum), which were pre-incubated for 24 h with 100 µg/mL of each AFP. Fruits were stored at 20 • C and 90% relative humidity. Each wound was scored daily for infection symptoms on consecutive days post-inoculation (dpi). Statistical analyses were performed using IBM SPSS Statistics v.26 to calculate one-way ANOVA and Tukey's HSD test (p < 0.05). Representative images of infected P. digitatum and P. italicum orange fruits and P. expansum apple fruits are shown in Figure 1c.

Antifungal Activity Assays
The three P. chrysogenum AFPs and N. fischeri NFAP2 were tested for their antifungal activity against the three main Penicillium species that cause severe postharvest fruit diseases (P. digitatum, P. italicum and P. expansum) (Figure 1a,c). Although previously characterized, PAF was included in the antifungal assays as an internal control for a better comparison among P. chrysogenum AFPs. Amino acid sequence alignment and physicochemical properties of the AFPs evaluated together with those of the highly active PeAfpA [13] and PdAfpB [14] are summarized in Figure 1b and Table 1. Table 2 shows MIC values of each AFP against P. digitatum, P. italicum and P. expansum. PAFB was the most potent AFP against the three Penicillium species, with MIC values similar to those described for PeAfpA and PdAfpB (Table 2) [13,14]. NFAP2 showed a moderate antifungal activity against P. digitatum and P. italicum, although no MIC value for P. expansum could be determined at the highest concentration tested (64 µM). Similarly, PAFC completely inhibited the growth of P. digitatum and P. italicum with MIC values of 4 µM and 8 µM, respectively, whereas the MIC value for P. expansum was reached at 32 µM. As expected, the three Penicillium species were moderately resistant to PAF, confirming previous results [14].

PAFB Delays P. digitatum Infection in Orange Fruits
Based on the in vitro antifungal susceptibility test results, experiments were designed to evaluate the ability of the two most active proteins, PAFB and NFAP2, to control the green mold disease caused by P. digitatum infection to citrus fruit. PeAfpA, which had been previously described as effective in the control of the fungus in orange fruits [13], was included as an internal control. Figure 2a shows the effects of 100 µg/mL AFPs (corresponds to 15 µM PeAfpA, 16 µM PAFB and 18 µM NFAP2) in orange fruits from the Navel variety. The three proteins showed a slight control of P. digitatum infection. PeAfpA delayed fungal infection throughout the experiment; PAFB did not affect fungal growth at 4 dpi, although it significantly decreased the incidence of infection at 5, 6 and 7 dpi, while NFAP2 failed to control P. digitatum at 7 dpi. The average efficacy at 7 dpi was around 30% and 20% of disease reduction for PeAfpA and PAFB, respectively. In order to confirm the effect of AFPs, a second experiment with AFPs in individual and combined treatments was carried out (Figure 2b). Both PeAfpA and PAFB treatments resulted in a delay of fungal infection, although the average efficacy of disease reduction at 7 dpi of PeAfpA (25%) and PAFB (40%) varied with respect to that of Figure 2a. Regarding NFAP2, the effect was only statistically significant at 3 dpi. With respect to AFP combinations, the combined effect of PeAfpA and PAFB was not different to that caused by individual proteins, suggesting no additive or synergistic effects. Furthermore, combinations of either PeAfpA or PAFB with NFAP2 did not result in any protective effect. Figure 2c shows representative images of AFP-treated oranges at 7 dpi.

PAFB Delays P. italicum Infection in Orange Fruits
Based on the individual MIC values against P. italicum, PAFB, PAFC and NFAP2 were selected to control the blue mold disease in orange fruits. PeAfpA, with a MIC value of 0.3 µM against this fungus (Table 2) [13], was included in the study since its in vivo effect has not been previously reported. Figure 3a shows the effect of the four proteins at 100 µg/mL (corresponds to 15 µM PeAfpA, PAFC and PdAfpB, 16 µM PAFB and 18 µM NFAP2). Only PAFB showed a slight reduction in the incidence of infection compared to the control with no AFP treatment. The PAFB average efficacy of blue mold reduction along the experiment varied from 60% at 4 dpi to 15% at 7 dpi. With the aim of confirming the efficacy of PAFB, a second infection experiment was designed including PdAfpB, another class B protein, whose MIC value against P. italicum was determined to be 0.3 µM (Table 2) [14]. Figure 3b shows the effect of both class B AFPs and PeAfpA on orange fruit infections from Lanelate variety. PAFB controlled the experimental P. italicum infection from 5 to 7 dpi while neither PdAfpB nor PeAfpA exhibited any infection inhibitory effect. Figure 3c shows representative images of AFP-treated Lanelate oranges at 6 dpi, where the average efficacy of PAFB was 50% disease reduction. At the end of the experiment (7 dpi), the average efficacy of PAFB treatment was 40% disease reduction. gi 2021, 7, x FOR PEER REVIEW 6 of 13

PAFB Delays P. expansum Infection in Apple Fruits
Previous in vivo experiments indicated that PeAfpA efficiently protected apple fruits against P. expansum infections [31]. Here, we conducted apple inoculation experiments to assess the effectiveness of the most active AFP in vitro, PAFB, in comparison to PeAfpA. Figure 4a shows the effect of both proteins at 100 µg/mL (corresponds to 15 µM PeAfpA and 16 µM PAFB) on the P. expansum infection of Golden Delicious apples. Both PeAfpA and PAFB controlled the infection with similar efficacy (Figure 4a). At 7 dpi, disease reductions close to 50% were observed. To confirm the control of P. expansum infection by PAFB treatment, a second independent experiment was accomplished (Figure 4b). PAFB exerted a significant protective effect, although the efficacy after 7 dpi was lower (32% of disease reduction) than that observed in Figure 4a. Combination of PAFB and PeAfpA at 15 µM each did not further improve the effect caused by the individual treatments.

PAFB Delays P. expansum Infection in Apple Fruits
Previous in vivo experiments indicated that PeAfpA efficiently protected apple fruits against P. expansum infections [31]. Here, we conducted apple inoculation experiments to assess the effectiveness of the most active AFP in vitro, PAFB, in comparison to PeAfpA. Figure 4a shows the effect of both proteins at 100 µg/mL (corresponds to 15 µM PeAfpA and 16 µM PAFB) on the P. expansum infection of Golden Delicious apples. Both PeAfpA and PAFB controlled the infection with similar efficacy (Figure 4a). At 7 dpi, disease reductions close to 50% were observed. To confirm the control of P. expansum infection by PAFB treatment, a second independent experiment was accomplished (Figure 4b). PAFB exerted a significant protective effect, although the efficacy after 7 dpi was lower (32% of disease reduction) than that observed in Figure 4a. Combination of PAFB and PeAfpA at 15 µM each did not further improve the effect caused by the individual treatments.

Discussion
Currently, Penicillium postharvest decay control strategies rely on the use of synthetic fungicides such as imazalil, sodium ortho-phenylphenate or thiabendazole, although extensive research for the development of new alternatives is being conducted [3,5]. Natural alternative methods include plant extracts, natural antifungal edible coatings and antifungal peptides and proteins [1] as those evaluated in this work.
In the present study, we describe the potential of several phylogenetically different AFPs as new agents for Penicillium decay control. The AFPs tested were PAF, PAFB and PAFC from P. chrysogenum, and NFAP2 from N. fischeri. Although antifungal activity of the four AFPs was previously tested against different microorganisms [15,27,30,34,36], the three main Penicillium species that cause significant economic losses after harvest, P. digitatum, P. italicum and P. expansum, were not included in those studies. The AFPs evaluated differ in amino acid composition, primary structure and physicochemical

Discussion
Currently, Penicillium postharvest decay control strategies rely on the use of synthetic fungicides such as imazalil, sodium ortho-phenylphenate or thiabendazole, although extensive research for the development of new alternatives is being conducted [3,5]. Natural alternative methods include plant extracts, natural antifungal edible coatings and antifungal peptides and proteins [1] as those evaluated in this work.
In the present study, we describe the potential of several phylogenetically different AFPs as new agents for Penicillium decay control. The AFPs tested were PAF, PAFB and PAFC from P. chrysogenum, and NFAP2 from N. fischeri. Although antifungal activity of the four AFPs was previously tested against different microorganisms [15,27,30,34,36], the three main Penicillium species that cause significant economic losses after harvest, P. digitatum, P. italicum and P. expansum, were not included in those studies. The AFPs evaluated differ in amino acid composition, primary structure and physicochemical properties (Figure 1 and Table 1), which might explain the different in vitro antifungal profiles and in vivo efficacy in fruit inoculation experiments observed in this study.
PAFB displayed the highest antifungal potency in in vitro assays with MIC values in the range of 0.12-0.25 µM, comparable to those of PeAfpA (0.15-0.3 µM) [13] and PdAfpB (0.3-0.6 µM) [14]. PAFB also exhibits growth inhibitory activity against human pathogenic fungi such as Aspergillus fumigatus, Trichophyton spp. and Candida spp., and against the fungal model organisms Neurospora crassa and Saccharomyces cerevisiae [15]. Remarkably, these latter species were similarly sensitive towards PAFB and PAF with MIC values in the range 0.25-4 µM [15,25], while our results show that the three phytopathogenic Penicillium species tested were much more resistant to PAF (MIC values in the range 16-32 µM) than to PAFB. These results show that AFPs act in a species-specific manner.
PAFC and NFAP2 showed a moderate antifungal activity against P. digitatum and P. italicum, whereas the MIC values against P. expansum were much higher. The in vitro growth inhibitory activity of PAFC against opportunistic human pathogenic members of the Candida genus has been recently described [30]. Remarkably, PAFC shares 83% amino acid identity with the P. expansum class C protein, PeAfpC, which did not show any antifungal activity against the three Penicillium species tested [13]. Only recently, the antifungal activity of PeAfpC against several species of the genus Byssochlamys has been reported [37]. Regarding NFAP2, it was characterized having a unique high anti-yeast activity whereas it was ineffective against filamentous fungi [27]. Here, our results extend the antifungal spectrum of both PAFC and NFAP2 to filamentous fungi of the Penicillium genus.
The AFP from Aspergillus giganteus was the first experimentally tested ascomycetous antifungal protein to control plant and postharvest diseases in vivo. A. giganteus AFP has been reported to successfully control postharvest decay caused by Magnaporthe oryzae in rice [38], Fusarium oxysporum in tomato plant seeds [39] and the infection caused by Alternaria alternata in banana [40]. Here, based on the in vitro inhibitory efficacy, we selected different AFPs to evaluate their potential in the control of Penicillium decay using three pathosystems, P. digitatum-orange fruit, P. italicum-orange fruit and P. expansum-apple fruit. PAFB provoked disease reductions in orange and apple fruits, although with moderate efficacy. In agreement to that observed in the in vitro assays, the PAFB effect in protection experiments was equivalent to that observed with PeAfpA in the side-by-side experiments conducted. We showed previously that PeAfpA exerts protection against P. digitatum in oranges and against P. expansum in apples at concentrations of 0.15-15 µM, although variations in the percentage of disease reduction among experiments, as those observed here, were reported [13,31]. PAFB was also effective in delaying P. italicum infection to Navel and Lanelate oranges while PeAfpA did not show any effect. This is the first time that PeAfpA is evaluated in the pathosystem P. italicum-orange fruit, and despite its potency against the fungus in in vitro experiments [13], no protection effect was observed in any of the orange varieties tested. Discrepancies between in vitro and in vivo experiments were also found for PAFC, NFAP2 and PdAfpB, since none of them were able to efficiently protect orange fruits despite their in vitro determined MIC values against P. digitatum and P. italicum (Table 2). We previously described that PAF, but not the rationally designed variant PAF opt , was able to inhibit Botrytis cinerea infection in tomato plant leaves, although both proteins inhibited B. cinerea growth in vitro [20]. In in vivo experiments additional factors that are absent in or differ from in vitro assays impact the antifungal potential of AFPs, e.g., fruit-specific substrates may influence fungal growth and AFP susceptibility, or compounds present in orange peels and apple skins may interfere with the AFP activity. Our study emphasizes the need for in vivo protection assays, as those described here, to evaluate the feasibility of AFPs in postharvest control.
PAFB is orthologous to PgAFP identified in the supernatant of P. chrysogenum strain RP42C, originally isolated from dry-cured ham [41]. PgAFP exhibits potent inhibitory activity against the main mycotoxin-producing species of Aspergillus and Penicillium of concern for dry-ripened foods and it efficiently reduces counts of Aspergillus flavus and Penicillium restrictum inoculated on a dry fermented sausage [42]. Recently, the effect of PgAFP against P. expansum and P. digitatum growth on fruits has been evaluated [43]. Although in vitro growth inhibition was shown, no inhibitory effect was observed in oranges from Navelina and Lanelate varieties or in Golden Delicious apples. However, a protective effect was found in apples from the Royal Gala variety [43]. It should be mentioned that it is difficult to compare inter-laboratory in vivo experiments mainly due to different protocols of AFP application and fungal inoculation. Moreover, the effects of the fungal strain as well as of the fruit variety should be considered.
Results reported here for AFPs only show a delay in disease progression, far from the level of protection attributed to chemical fungicides. However, the conditions of our controlled inoculation experiments should be considered. AFPs are point-inoculated and mycelia growing out from the inoculation site (where the protein is absent) might contribute to disease incidence, while fungicides are commercially applied onto the entire fruit surface. In addition, the inoculum dose used in protection assays is aggressive since it renders around 80-100% of infection at 6-7 dpi. Interestingly, when fungicides such as imazalil and thiabendazole are applied in parallel assays with antifungal peptides under similar conditions as those described here, fungicides performed similarly and the average efficacy of disease reduction was not significantly different to that provoked by AFPs or antifungal peptides [44,45]. Further experiments mimicking commercial conditions of fungicide application are necessary to confirm the feasibility of AFPs in postharvest protection.
To reach the level of efficacy provided by conventional fungicides, combinations of alternative approaches of the same or different nature have been proposed [1]. Here, we have combined PeAfpA and PAFB, the two active AFPs in fruit protection experiments, to evaluate a potential additive or synergistic effect. However, the combination of both proteins in the control of P. digitatum infection in orange fruits ( Figure 2) and P. expansum infection in apples (Figure 4), respectively, did not result in a synergistic effect and not even an additive effect was observed. To the best of our knowledge, this is the first time that a combination of AFPs has been reported in in vivo experiments, although under the conditions tested it did not improve the efficacy observed with the individual treatments. Notably, when either PAFB or PeAfpA were combined with NFAP2 (Figure 2c), the latter counteracted the effect of both proteins, suggesting a negative interaction between these AFPs. Nevertheless, optimization of concentrations and ratios at which AFPs should be combined require further research. In addition, a precise understanding of the mechanism of synergistic interaction of antimicrobials is necessary for the design of combined approaches. In this context, the mechanism of antifungal action of AFPs, as described for PAF, PAFB and PdAfpB, is complex and regulated [25,46]. These three proteins have an energy-dependent cell-penetrating mode of action followed by a series of intracellular regulated actions that end with cell collapse [15,25,46,47]. The PAFC mode of action also requires PAFC uptake and cytoplasmic localization before plasma permeabilization occurs [30]. In contrast, this mechanism differs from that of NFAP2, whose cell-killing activity seems connected to its pore-forming ability in the cell membrane [16]. Although the mechanism of PeAfpA is still unknown, our data suggest that the combination of two AFPs with different mechanisms of action as those of PAFB and NFAP2 do not result in an increase in the efficacy. Further studies are necessary to determine the PeAfpA mode of action as well as to unravel the potential interactions among AFPs. Combinations of AFPs with chemical fungicides or physical treatments are in progress.

Conclusions
This study provides additional knowledge about the three P. chrysogenum AFPs and N. fischeri NFAP2, which show a species-specific inhibition spectrum against the main Penicillium species causing postharvest decay. PAFB, the most active antifungal protein in in vitro susceptibility tests, delays P. digitatum, P. italicum and P. expansum infection in