Diversity of Mycotoxins and Other Secondary Metabolites Recovered from Blood Oranges Infected by Colletotrichum, Alternaria, and Penicillium Species

This study identified secondary metabolites produced by Alternaria alternata, Colletotrichum gloeosporioides, and Penicillium digitatum in fruits of two blood orange cultivars before harvest. Analysis was performed by UHPLC–Q-TOF-MS. Three types of fruits were selected, asymptomatic, symptomatic showing necrotic lesions caused by hail, and mummified. Extracts from peel and juice were analyzed separately. Penicillium digitatum was the prevalent species recovered from mummified and hail-injured fruits. Among 47 secondary metabolites identified, 16, 18, and 13 were of A. alternata, C. gloeosporioides, and P. digitatum, respectively. Consistently with isolations, indicating the presence of these fungi also in asymptomatic fruits, the metabolic profiles of the peel of hail-injured and asymptomatic fruits did not differ substantially. Major differences were found in the profiles of juice from hail-injured and mummified fruits, such as a significant higher presence of 5,4-dihydroxy-3,7,8-trimethoxy-6C-methylflavone and Atrovenetin, particularly in the juice of mummified fruits of the Tarocco Lempso cultivar. Moreover, the mycotoxins patulin and Rubratoxin B were detected exclusively in mummified fruits. Patulin was detected in both the juice and peel, with a higher relative abundance in the juice, while Rubratoxin B was detected only in the juice. These findings provide basic information for evaluating and preventing the risk of contamination by mycotoxins in the citrus fresh fruit supply chain and juice industry.

The aims of this study were (i) to apply the UHPLC-Q-TOF-MS analytical method for the characterization of the mycotoxicological profile of mature fruits of two blood orange cultivars, (ii) to determine how this profile varies depending on the fungi associated with the fruit peel and in the diverse parts of the fruit, and (iii) to investigate how the profile is affected by the phytosanitary status of the fruit as a consequence of an environmental stress, such as hail.

Results
Approximately 2800 fungal mass isolates were obtained from orange fruits. Mass isolates were preliminarily separated into three groups on the basis of morphotype, i.e., colony morphology on PDA and microscopical traits. The three morphotypes corresponded to Alternaria, Colletotrichum, and Penicillium, respectively. On PDA, Alternaria isolates showed flat, woolly colonies ranging from dark green to black in color. They produced typical dark brown, club-shaped conidia arranged in branched chains, with oval-ellipsoidal shapes and 3-5 transverse septa. Colonies of Colletotrichum isolates grown on PDA showed a dense, cottony, aerial mycelium that was initially white and turned progressively pale gray with salmon-pink conidial mucilaginous masses in the center of the colony. Dark acervuli were scattered over the entire surface of old colonies, and the colony reverse was pale orange to uniformly grey. The single-celled conidia were hyaline, smooth, cylindrical, with both ends rounded, and dimensions of 11-15 × 4-6 µm. Setae were present in most isolates. On PDA, colonies of Penicillium isolates were initially white and then turned green due to abundant production of conidia. The colony margin was entire and narrow, and conidia were borne in chain on typically branched conidiophores; they were globose to subglobose in shape, smooth, with a size range of 2.51-4.22 × 2.35-3.61 µm and mean size (±SD) of 3.38 ± 0.49 × 3 ± 0.36 µm.
Isolates of Alternaria, Colletotrichum, and Penicillium accounted for 12, 28, and 60% of all isolates recovered from orange fruits, respectively, and the proportions of isolates of the three fungal genera in each of the six clusters of fruits are reported in Table 1. A set of about 7% of mass isolates of each fungal genus, comprising isolates from all clusters, was randomly selected, and a single-conidium isolate was obtained from each selected mass isolate to be identified molecularly. Overall, 200 isolates, 24 of Alternaria, 56 of Colletotrichum. and 120 of Penicillium, were identified molecularly at species level. In particular, based on the phylogenetic analysis of both the ITS and β-tub regions, the Colletotrichum and Penicillium isolates were identified as C. gloeosporioides and P. digitatum, respectively, while based on the phylogenetic analysis of ITS, TEF-1a, and Alt-1a regions, the Alternaria isolates were identified as A. alternata. As the isolates belonging to the same species were as a result identical, a limited number of sequences of isolates per each species were deposited on Genbank ( Table 2). Colletotrichum gloeosporioides OQ929085 OQ938693 --ID1O20 Colletotrichum gloeosporioides Colletotrichum gloeosporioides OQ929097 OQ938705 -- The analysis by Agilent Ultra High-Definition Accurate Mass revealed 47 diverse fungal secondary metabolites in peel and juice of orange fruits. The identification of the metabolites was supported by using a specifically designed database based on the secondary metabolites produced by species of Alternaria, Colletotrichum, and Penicillium available in the literature (Table 3).  For the calibration of the UHPLC-Q-TOF-MS analysis method, commercial mycotoxins (Sigma-Aldrich) of two of the identified fungi were used as standards, patulin for P. digitatum and AME for A. alternata, respectively. Precision was performed by inter-and intraday variations to make the analytical method accurate. Mean recoveries were operated on orange juice and orange peel (non-contaminated) using three repetitive injections of selected standards in a day and three consecutive days for the interday variation at levels of patulin and AME (0.01, 0.1 and 1 mg/L for both toxins) which were of 81.1 ± 1.9% and 89.4 ± 2.4% for orange juice and 84.9 ± 2.6% and 92.6 ± 3.1% for orange peel. The results were expressed as relative standard deviations (RSD) of intra-day and interday variation with values ranging between 3.7 and 4.1% for patulin and between 2.9 and 4.5% for AME, respectively. The matrix effects (% MEs) evidenced for patulin and AME in orange juice were 89% and 92%, whereas in orange peel the percentages of the MEs detected were 96% and 98%, respectively. The secondary metabolites detected by UHPLC-Q-TOF-MS in each fruit cluster are presented as heat maps (Figures 1-3). Only secondary metabolites with a large gap between the asymptomatic fruit cluster of each orange cultivar and the clusters of both hail-injured and mummified fruits of the same cultivar are reported in the maps.
of selected standards in a day and three consecutive days for the interday variation at levels of patulin and AME (0.01, 0.1 and 1 mg/L for both toxins) which were of 81.1 ± 1.9% and 89.4 ± 2.4% for orange juice and 84.9 ± 2.6% and 92.6 ± 3.1% for orange peel. The results were expressed as relative standard deviations (RSD) of intra-day and interday variation with values ranging between 3.7 and 4.1% for patulin and between 2.9 and 4.5% for AME, respectively. The matrix effects (% MEs) evidenced for patulin and AME in orange juice were 89% and 92%, whereas in orange peel the percentages of the MEs detected were 96% and 98%, respectively.
The secondary metabolites detected by UHPLC-Q-TOF-MS in each fruit cluster are presented as heat maps (Figures 1-3). Only secondary metabolites with a large gap between the asymptomatic fruit cluster of each orange cultivar and the clusters of both hailinjured and mummified fruits of the same cultivar are reported in the maps.     Figure 1A shows that the secondary metabolites of A. alternata present in orange peel included SM2, SM3, SM4, SM5, SM6, SM9, SM10, SM11, SM12, and SM15. However, only SM3 (AF-toxins II) and SM6 (Alternethanoxin A) were found with a high relative abundance compared to the other metabolites. Specifically, SM3 was found with a high relative abundance in all clusters except for the mummified fruits of both varieties, while SM6 was relatively abundant in all clusters, except in Mum_L.
The secondary metabolites detected in orange juice using the Penicillium dataset were SM35, SM38, SM39, SM40, SM41, SM43, SM44, SM45, and SM47. Specifically, SM35 (Alantrypinone) and SM38 (Atrovenetins) were found with a higher relative abundance compared to the other metabolites. In juice of mummified fruits of both orange cultivars, the (Mum_L) and (Mum_T) clusters, the relative abundance of SM35 (Alantrypinone) was approximately four-and three-fold higher than in juice of asymptomatic fruits of both cultivars, the (Asym_L) and (Asym_T) clusters, respectively. Similarly, the relative abundance of SM38 (Atrovenetins) in mummified fruits of both orange cultivars, the (Mum_L) and (Mum_T) clusters, was approximately four-and three-fold higher than in asymptomatic fruits, the (Asym_L) and (Asym_T) clusters, respectively. In the (Mum_L) cluster, SM39 (Fungisporin) showed a relative abundance approximately 355-fold higher than in cluster (Asym_L); it was present in the (Mum_T) cluster but was not detected in cluster (Asym_T). SM40 (Lichexanthone) was found in cluster (Mum_L) but not in cluster (Asym_T). Moreover, this metabolite in (Mum_T) showed a relative abundance approximately 13-fold higher than in cluster (Asym_T). SM43 (Penochalasin K), in mummified fruits of 'Tarocco Lempso', cluster (Mum_L), showed a relative abundance approximately 11-fold higher than in asymptomatic fruits of the same cultivar, cluster (Asym_L). This metabolite was also detected in mummified fruits of 'Tarocco Tapi', cluster (Mum_T), but was not found in asymptomatic fruits of this orange cultivar, cluster (Asym_T). SM44 (Rubratoxin B) and SM47 (patulin) were found only in the (Mum_L) and (Mum_T) clusters. In mummified fruits of the two orange cultivars, clusters (Mum_L) and (Mum_T), SM45 (Serantrypinone) was detected with a relative abundance about five-and four-fold higher than in asymptomatic fruits of the respective cultivars, clusters (Asym_L) and (Asym_T).
Principal Component Analysis (PCA) of data was performed. It was based on the secondary metabolites of A. alternata, C. gloeosporioides, and P. digitatum (Table 2) identified in the peel and juice of asymptomatic, hail-injured, and mummified fruits of the two blood orange cultivars, 'Tarocco Lempso' and 'Tarocco Tapi'. Figure 4 shows the clustering of asymptomatic, hail-injured, and mummified orange fruits of the two orange cultivars in the score plot and loading plot, based on secondary metabolites of Alternaria detected in the peel. The sum of the principal component values accounted for 69.9% of the total variance, with PC1 representing 47.3% and PC2 representing 22.6% of the total variance.
In the score plot, asymptomatic orange fruits (Asym_T and Asym_L) were clustered in quadrants II and IV, while hail-injured fruits (H-Inj_T and H-Inj_L) were clustered in quadrant II. Mummified fruits of 'Tarocco Lempso' (Mum_L), with the only exception of replicate #1, tended towards quadrant IV. Conversely, mummified fruits of 'Tarocco Tapi' (Mum_T) were clustered in quadrant III.
The score plot and loading plot of Figure 5 illustrate the clustering of the three types of fruits (asymptomatic, hail-injured, and mummified) of 'Tarocco Lempso' and 'Tarocco Tapi' based on secondary metabolites of Colletotrichum detected in the peel. The principal component values accounted for 72.2% of the total variance, with PC1 and PC2 accounting for 58.6% and 13.6%, respectively. metabolites of Alternaria detected in the peel. The sum of the principal component values accounted for 69.9% of the total variance, with PC1 representing 47.3% and PC2 representing 22.6% of the total variance. In the score plot, asymptomatic orange fruits (Asym_T and Asym_L) were clustered in quadrants II and IV, while hail-injured fruits (H-Inj_T and H-Inj_L) were clustered in quadrant II. Mummified fruits of 'Tarocco Lempso' (Mum_L), with the only exception of replicate #1, tended towards quadrant IV. Conversely, mummified fruits of 'Tarocco Tapi' (Mum_T) were clustered in quadrant III.
The score plot and loading plot of Figure 5 illustrate the clustering of the three types of fruits (asymptomatic, hail-injured, and mummified) of 'Tarocco Lempso' and 'Tarocco Tapi' based on secondary metabolites of Colletotrichum detected in the peel. The principal component values accounted for 72.2% of the total variance, with PC1 and PC2 accounting for 58.6% and 13.6%, respectively.  In the score plot, the asymptomatic fruits of the two orange cultivars (Asym_L and Asym_T) were grouped in quadrant II and IV, respectively. The hail-injured fruits of 'Tarocco Lempso' (H-Inj_L) were clustered in quadrant III, while the hail-injured fruits of 'Tarocco Tapi' (H-Inj_T) were spread across quadrants I, II, and IV. The mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi' (Mum_L and Mum_T) were clustered in quadrants III and I, respectively, with the only exception of Mum_T replicate #1, which tended towards quadrant IV.
The loading plot showed that secondary metabolites associated with the peel of In the score plot, the asymptomatic fruits of the two orange cultivars (Asym_L and Asym_T) were grouped in quadrant II and IV, respectively. The hail-injured fruits of 'Tarocco Lempso' (H-Inj_L) were clustered in quadrant III, while the hail-injured fruits of 'Tarocco Tapi' (H-Inj_T) were spread across quadrants I, II, and IV. The mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi' (Mum_L and Mum_T) were clustered in quadrants III and I, respectively, with the only exception of Mum_T replicate #1, which tended towards quadrant IV.
In the loading plot, secondary metabolites SM19, SM22, SM24, SM27, SM28, SM30, SM31, SM32, and SM33 tended towards quadrants I and III, clustering the hail-injured fruits of 'Tarocco Lempso' (H-Inj_L) as well as the mummified fruits of both orange cultivars (Mum_L and Mum_T) ( Figure 5). Figure 6 shows the clustering of asymptomatic, hail-injured, and mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi' on the basis of Penicillium secondary metabolites detected in the peel (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV); the sum of the principal component values accounted for 69.7% of the total variance. PC1 represented 51.1% and PC2 18.6% of the total variance. The score plot of Penicillium metabolites detected in the peel shows that asymptomatic fruits (Asym_L and Asym_T) were clustered in quadrant IV, with the exception of replicate #3, which tended towards quadrant II. Hail-Injured fruits of 'Tarocco Lempso' (H-Inj_L) were clustered in quadrant I, while fruits of 'Tarocco Tapi' (H-Inj_T) were scattered in quadrants I, II, and IV. The mummified fruits of both cultivars (Mum_T and Mum_L) were clustered in quadrant III, with the exception of cluster (Mum_T) replicate # 1, which tended towards quadrant IV.
In the loading plot, secondary metabolites SM35, SM36, SM37, SM39, SM40, SM42, SM43, SM44, SM45, SM46, and SM47 tended towards quadrants I and III, clustering the samples of peel of hail-injured fruits of 'Tarocco Lempso' (H-Inj_L), replicate #3 of hail-injured fruits of 'Tarocco Tapi' (H_Inj_T_3), and mummified fruits of both orange cultivars (Mum_L and Mum_T) ( Figure 6). Figure 7 shows the clustering of asymptomatic, hail-injured, and mummified orange fruits of 'Tarocco Lempso' and 'Tarocco Tapi' based on Alternaria metabolites (score plot) detected in the juice and the metabolite trends (loading plot) in different quadrants (I, II In the loading plot, the secondary metabolites SM1, SM7, SM8, SM13, and SM16 tended towards quadrant II, clustering the mummified fruits of 'Tarocco Lempso' (Mum_L and Mum_T), and the metabolites SM2 SM4, SM5, SM6, SM10, and SM14 clustered the hail-injured fruits of both orange cultivars (H-Inj_L and Mum_T) (Figure 7). Figure 8 shows the clustering of asymptomatic, hail-injured, and mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi', based on Colletotrichum secondary metabolites detected in the juice (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV). The sum of the principal 14 component values accounted for 70.3% of the total variance. PC1 and PC2 accounted for 50.9 and 19.4% of the total variance, respectively. In the score plot of Alternaria alternata, asymptomatic fruits of the two orange cultivars (Asym_T and Asym_L) were clustered in quadrants I and III, respectively, with the only exception of replicate #3 of asymptomatic fruits of 'Tarocco Tapi' (Asym_T_3), which tended towards quadrant III. Hail-injured fruits of 'Tarocco Lempso' (H-Inj_L) were clustered in quadrant III, while hail-injured fruits of 'Tarocco Tapi' (H-Inj_T) were clustered in quadrant I, with the only exception of replicate #3 (H-Inj_T_3) that tended towards quadrant III. Mummified fruits of 'Tarocco Lempso' (Mum_L) were clustered in quadrant II, while mummified fruits of 'Tarocco Tapi' (Mum_T) were clustered in quadrants I, II, and IV.
In the loading plot, the secondary metabolites SM1, SM7, SM8, SM13, and SM16 tended towards quadrant II, clustering the mummified fruits of 'Tarocco Lempso' (Mum_L and Mum_T), and the metabolites SM2 SM4, SM5, SM6, SM10, and SM14 clustered the hail-injured fruits of both orange cultivars (H-Inj_L and Mum_T) (Figure 7). Figure 8 shows the clustering of asymptomatic, hail-injured, and mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi', based on Colletotrichum secondary metabolites detected in the juice (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV). The sum of the principal 14 component values accounted for 70.3% of the total variance. PC1 and PC2 accounted for 50.9 and 19.4% of the total variance, respectively. Figure 8 shows the clustering of asymptomatic, hail-injured, and mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi', based on Colletotrichum secondary metabolites detected in the juice (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV). The sum of the principal 14 component values accounted for 70.3% of the total variance. PC1 and PC2 accounted for 50.9 and 19.4% of the total variance, respectively.  In the score plot of Colletotrichum gloeosporioides, asymptomatic fruits of both orange cultivars (Asym_T and Asym_L) were clustered in quadrants II and IV, respectively, with the only exception of replicate #3 of 'Tarocco Tapi' (Asym_T_3), which tended towards quadrant IV. Hail-injured fruits of both orange cultivars (H-Inj_L and H-Inj_T) were clustered in quadrant III, with the only exception of replicate #1 of hail-injured 'Tarocco Tapi' (H-Inj_T_1), which tended towards quadrant II. Mummified fruits of both orange cultivars (Mum_L and Mum_T) were clustered in quadrant I, with the only exception of replicate #1 of mummified 'Tarocco Tapi' (Mum_T_1), which tended towards quadrant II. In the loading plot, the secondary metabolites, SM18, SM19, SM23, SM29, SM30, and SM34, tended towards quadrant I, clustering the mummified fruits of both orange cultivars (Mum_L and Mum_T), while the secondary metabolites SM17, SM20, SM24, SM25, SM26, and SM32 tended towards quadrant III, clustering the hail-injured fruits of both orange cultivars (H-Inj_L and H-Inj_T) (Figure 8). Figure 9 shows the clustering of asymptomatic, hail-injured, and mummified fruits of 'Tarocco Lempso' and 'Tarocco Tapi', based on Penicillium secondary metabolites detected in the juice (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV). The sum of the principal nine component values accounted for 77.2% of the total variance. PC1 and PC2 accounted for 60.2 and 17% of the total variance, respectively.
In the score plot of Penicillium digitatum, asymptomatic fruits of the two orange cultivars (Asym_T and Asym_L) were clustered in quadrants I and III. Hail-injured fruits of both orange cultivars (H-Inj_L and H-Inj_T) were clustered in quadrants I and II, while mummified fruits of the two cultivars (Mum_L and Mum_T) were clustered in quadrants III and IV. In the loading plot, the secondary metabolites SM38, SM39, SM41, and SM45 tended towards quadrant II, clustering the hail-injured fruits of both 'Tarocco Lempso' and 'Tarocco Tapi' (H-Inj_L and H-Inj_T), while the secondary metabolites SM40 and SM47 tended towards quadrant IV, clustering the mummified fruits of both orange cultivars (Mum_L and Mum_T) (Figure 9). of 'Tarocco Lempso' and 'Tarocco Tapi', based on Penicillium secondary metabolites detected in the juice (score plot) and the metabolite trends (loading plot) in different quadrants (I, II, III, and IV). The sum of the principal nine component values accounted for 77.2% of the total variance. PC1 and PC2 accounted for 60.2 and 17% of the total variance, respectively. In the score plot of Penicillium digitatum, asymptomatic fruits of the two orange cultivars (Asym_T and Asym_L) were clustered in quadrants I and III. Hail-injured fruits of both orange cultivars (H-Inj_L and H-Inj_T) were clustered in quadrants I and II, while mummified fruits of the two cultivars (Mum_L and Mum_T) were clustered in quadrants III and IV. In the loading plot, the secondary metabolites SM38, SM39, SM41, and SM45 tended towards quadrant II, clustering the hail-injured fruits of both 'Tarocco Lempso' and 'Tarocco Tapi' (H-Inj_L and H-Inj_T), while the secondary metabolites SM40 and SM47 tended towards quadrant IV, clustering the mummified fruits of both orange cultivars (Mum_L and Mum_T) (Figure 9).

Discussion
In this study, isolations with a conventional microbiological method from the fruit peel of two blood orange cultivars, 'Tarocco Lempso' and 'Tarocco Tapi', sampled in commercial citrus orchards of a typical production area of eastern Sicily, yielded three species of fungi, A. alternata, C. gloeosporioides, and P. digitatum. Fruits of each cultivar were separated into three distinct groups on the basis of external symptoms, i.e., fruits with the rind injured by the hail, mummified fruits, and asymptomatic fruits, with no visible lesions of the rind. Then, secondary metabolites of peels and juices from the three groups of fruits were extracted and analyzed by UHPLC-Q-TOF-MS. Penicillium digitatum was the prevalent species recovered from mummified and hail-injured fruits. The UHPLC-Q-TOF-MS analysis revealed 47 secondary metabolites, of which 16, 18, and 13 were specifically associated with A. alternata, C. gloeosporioides, and P. digitatum, respectively. As regards the metabolites recovered from the three groups of fruits, major differences were found in the profiles of juice from hail-injured and mummified fruits, with a significant higher presence of 5,4-dihydroxy-3,7,8-trimethoxy-6C-methylflavone and Atrovenetin in the juice of

Discussion
In this study, isolations with a conventional microbiological method from the fruit peel of two blood orange cultivars, 'Tarocco Lempso' and 'Tarocco Tapi', sampled in commercial citrus orchards of a typical production area of eastern Sicily, yielded three species of fungi, A. alternata, C. gloeosporioides, and P. digitatum. Fruits of each cultivar were separated into three distinct groups on the basis of external symptoms, i.e., fruits with the rind injured by the hail, mummified fruits, and asymptomatic fruits, with no visible lesions of the rind. Then, secondary metabolites of peels and juices from the three groups of fruits were extracted and analyzed by UHPLC-Q-TOF-MS. Penicillium digitatum was the prevalent species recovered from mummified and hail-injured fruits. The UHPLC-Q-TOF-MS analysis revealed 47 secondary metabolites, of which 16, 18, and 13 were specifically associated with A. alternata, C. gloeosporioides, and P. digitatum, respectively. As regards the metabolites recovered from the three groups of fruits, major differences were found in the profiles of juice from hail-injured and mummified fruits, with a significant higher presence of 5,4-dihydroxy-3,7,8-trimethoxy-6C-methylflavone and Atrovenetin in the juice of mummified fruits of the Tarocco Lempso cultivar. Moreover, the mycotoxins patulin and Rubratoxin B were detected exclusively in mummified fruits. Patulin was detected in both the juice and peel, with a higher relative abundance in the juice, while Rubratoxin B was detected only in the juice.
The prevalent isolation of Penicillium digitatum from both mummified and hail-injured fruits of both orange cultivars suggests this fungus had a prominent role in fruit mummification and confirms its nature of being a wound parasite. Both C. gloeosporioides and A. alternata were recovered from necrotic lesions of hail-injured fruits of both orange cultivars in a significant higher frequency than from the intact rind of asymptomatic fruits. These fungi are ubiquitous in citrus orchards and normally reside in the peel of citrus fruits as epiphytes, endophytes, or latent pathogens. Very probably, in fruits injured by hail, their populations increased because they colonized the rind lesions as opportunistic necrotrophic pathogens. A possible explanation of both the failure in isolating A. alternata and the low proportion of successful isolations of C. gloeosporioides from mummified fruits is that P. digitatum exerted an antagonistic activity against the other two fungal species, based on the competition for the substrate or the production of fungitoxic secondary metabolites by this fungus. Alternaria alternata, C. gloeosporioides, and P. digitatum are responsible for some of the most common and severe pre-and post-harvest diseases of citrus fruits worldwide [1,13,23]. Moreover, they belong to three among the most known genera of mycotoxigenic fungi [58,124]. Overall, Alternaria species produce more than 70 mycotoxins sensu lato [94], and more than 100 secondary metabolites of Colletotrichum species have been reported so far [103]. Penicillium species produce a wide range of mycotoxins of concern for human and animal health. Among mycotoxins occurring in foods and feeds contaminated by Penicillium species, the most important are ochratoxin A and patulin, which are regulated in a number of countries, and to a lesser extent cyclopiazonic acid [124]. The identification of culturable fungi recovered from the peel of 'Tarocco Lempso' and 'Tarocco Tapi' fruits oriented the design and development of Alternaria, Colletotrichum, and Penicillium metabolite databases, that in this study were essential tools for the exploitation of UHPLC-Q-TOF-MS analysis method. Very recently, this innovative analytical approach has been successfully used for identifying the secondary metabolites produced by diverse Colletotrichum species both in vitro and on olives infected by anthracnose [125]. Actually, in the present study the combined application of UHPLC-Q-TOF-MS and databases of fungal metabolites proved to be very effective in identifying the secondary metabolites released by Alternaria, Colletotrichum, and Penicillium in the peel and juice of fruits of 'Tarocco Lempso' and 'Tarocco Tapi'. Overall, as many as 47 diverse fungal secondary metabolites were identified in the peel and juice of fruits of these two blood orange cultivars. Of these metabolites, 16 were produced by A. alternata; 18 were produced by C. gloeosporioides, and 13 were produced by P. digitatum. According to the scientific literature, all these metabolites are functionally active, and most of them show diverse types of activity. However, not all bioactive secondary metabolites produced by A. alternata, C. gloeosporioides, and P. digitatum detected in the peel and juice of citrus fruits in this study can be regarded as mycotoxins, sensu stricto [28,126].
In particular, as for the metabolites of Alternaria, Alternariol monomethyl ether (AME) is a mycotoxin that can cause kidney damage and immunosuppression in humans and animals [40,102,127]. AME was also reported to induce mitochondrial apoptosis in human colon carcinoma cells [90]. Alternethanoxin A displayed an in vitro inhibitory activity of the growth of cancer cells [123]. Aurasperone-C is known to cause nervous system dysfunctions [128]. Curvularin, which in this study was found only in mummified fruits, was reported to be a carcinogenic and immunotoxic substance [129]. Altenusin (ALN) was reported to be a multifunctional metabolite, with antibacterial, antifungal, and antiparasitic activity [130]. The amount of this metabolite in orange fruits varied considerably. In juice of hail-injured fruits, its relative abundance was 3-to 6-fold higher than in asymptomatic fruits while in the juice of mummified fruits its relative abundance was even 63to 251-fold higher than in asymptomatic fruits. AAL-toxins were demonstrated to have phytotoxic, cytotoxic, and genotoxic activity [88,131]. AAL-toxin TE2 was found with a high relative abundance in the juice of mummified orange fruits of both cultivars. This metabolite belongs to the AAL-toxins, a group of mycotoxins produced by A. alternata structurally related to Fuminosins produced by Fusarium species. AAL-toxins were reported as contaminants of various crops and as host specific phytoxins (HSTs) [94,132]. In the A. alternata/tomato pathosystem, they have a key role as pathogenicity factors [88,133,134]. AF-toxin II is a mixture of three stereoisomers reported to be phytotoxic on strawberry and pear fruits [134]. In this study, AF-toxin II is reported for the first time on citrus fruits infected by A. alternata. This metabolite was detected only in the peel of asymptomatic and hail-injured fruits of both orange cultivars examined. Maculosin was proposed as a potential eco-friendly agrochemical with herbicidal activity [94], while Porriolide was shown to exert a significant antifungal activity [135]. In this study, Alternethanoxin A was found with a high relative abundance in all types of orange fruits, including asymptomatic, hail-injured, and mummified fruits. In the original description of this metabolite isolated from A. sonchi cultures, it was envisaged the possibility of using it as a mycoerbicide in view of its broad-spectrum phytotoxic activity [136].
As for Colletotrichum metabolites identified in this study, Colletofragarone A1, Colletoic acid, and Pyrenocine A were reported to have phytotoxic and, in a few cases, cytotoxic properties [137,138]. Other compounds such as 5,4-dihydroxy-3,7,8-trimethoxy-6Cmethylflavone and Apigenin-8-C-β-D-glucopyranoside, which in this study were found with a high relative abundance in all types of orange fruits examined, were reported in the literature as bioactive metabolites [101,103]. Apigenin, in particular, was demonstrated to have a broad-spectrum anticancer activity [103,139]. However, none of these Colletotrichum metabolites have been so far reported as a mycotoxin of concern for human and animal health.
As for Penicillium, it is known this fungus produces a wide range of secondary bioactive metabolites, including metabolites with cytotoxic activity such as Penicillic acid, patulin, and Ochratoxin A [124,140]. Patulin is a mycotoxin that has been found as a contaminant in various types of fruits, including oranges [54]. It has been identified as a potential hazard for human health and food safety [18]. Several studies evidenced patulin can cause adverse health effects in humans, including cytotoxicity, genotoxicity, and immunotoxicity [42,141]. The European Union and the United States Food and Drug Administration have set maximum limits for patulin in food products, including oranges [26].
In this study, consistently with the results of isolations, high levels of patulin were detected in mummified fruits of both orange cultivars, in particular in juice. Rubratoxin B was another secondary metabolite found with a high relative abundance in the juice of mummified orange fruits of both 'Tarocco Lempso' and 'Tarocco Tapi'. It is produced by several species of Penicillium [142]. Rubratoxin B is a mycotoxin exhibiting a range of acute and chronic toxic effects, including hepatotoxicity, nephrotoxicity, splenotoxicity, and genotoxicity. It is commonly found as contaminant in cereals, such as rice and wheat, and poses a potential health risk to humans and animals [126,143]. Lichexanthone is an additional secondary metabolite found in a considerable amount in the juice of mummified orange fruits. This is a well-known metabolite of lichens and was also found in many filamentous fungi, including endophytic Penicillium and Trichoderma species [144]. It has been reported to possess antimicrobial and antitumoral activity. Other secondary metabolites of Penicillium, recovered in orange fruits, included Alantrypinone, a metabolite with herbicidal activity [145]; Serantrypinone, a metabolite with insecticidal activity [146]; Atrovenetin, known for its antioxidant effects [114]; and Asteltoxin and Penochalasin K, both with proven antifungal activity [114,147]. Other Penicillium secondary metabolites with antimicrobial and antifungal activity such as Fungisporin, Palitantin, and Penipacid B were found in orange fruits. These metabolites were detected in both mummified and hail-injured fruits, and their presence seems correlated with the prevalence of P. digitatum over the other fungal species as inferred from the results of isolation. From an analogy with other plant pathogenic fungal species [148], it can be speculated that the ability of P. digitatum to produce several metabolites, which show allelopathic, antifungal, and antibacterial activity, might enhance the ecological fitness of this fungus. Moreover, this feature could explain the ability of P. digitatum to exclude or inhibit other less competitive species, such as A. alternata and C. gloeosporioides, from mummified fruits or to overgrow them during isolation of axenic cultures.
The PCA of the interactions among driving factors shaping the profile of secondary metabolites in orange fruits showed the profiles of mummified fruits were distinct from the profiles of either asymptomatic or hail-injured fruits. It is noteworthy the mycotoxin patulin was a characteristic component of the Penicillium metabolite profile of both the peel and juice of mummified fruits of both orange cultivars examined while the other Penicilliumrelated mycotoxin Rubratoxin B was detected with a high relative abundance only in the juice of this type of fruits. Moreover, the Penicillium metabolite profile from the peel of hailinjured fruits was distinct from the profile of asymptomatic fruits and was characterized by the presence of secondary metabolites, such as Alantrypinone, Anacine, Lichexanthone, Penipacid B, Serantrypinone, and Solistatine, whose toxic activity for humans and animals has been so far little investigated.
As for C. gloeosporioides, the metabolic profiles in both the peel and juice of hail-injured fruits of 'Tarocco Lempso' and 'Tarocco Tapi' grouped separately from the profiles of the corresponding asymptomatic fruits, suggesting an interaction between hail injuries and the biology of this fungus which is a common resident in the fruit peel but also a latent pathogen.
As for A. alternata, all the secondary metabolites identified in this study were recovered from mummified and hail-injured fruits. Only one of them, AME, has been reported as a mycotoxin of concern for human health produced by Alternaria species associated with citrus [24]. Little is known about the toxicological potential of the other Alternaria-related metabolites recovered from orange fruits in this study or the risk of fruit contamination. Most of the studies that were aimed to characterize the biological activity of these metabolites or the ability of fungal isolates to produce them were performed in vitro. In general, to evaluate the risk of food and feed contamination, it is noteworthy that overall 26 out of 47 metabolites identified in this study were recovered from both the fruit peel and the juice (6, 9, and 8 metabolites of A. alternata, C. gloeosporioides, and P. digitatum, respectively), 12 were recovered exclusively from the peel (4 per each of the three fungal species), and 12 were recovered exclusively from the juice (6 and 5 of A. alternata and C. gloeosporioides, respectively, 1 of P. digitatum). Consequently, the risk of contamination concerns both fresh fruits and processed products such as juices, marmalade, and citrus pulp, the last one used largely as a feed for cattle.

Conclusions
Diverse secondary metabolites produced by three among the most common fungi responsible for pre-and post-harvest decays of citrus fruits were extracted directly from fruits sampled in commercial citrus orchards. They were identified using the modern analytical method of UHPLC-Q-TOF-MS. Many of these metabolites were isolated for the first time from citrus. The medium-and long-term objectives of this study were to evaluate the risk of contamination by mycotoxins and to provide basic information for setting maximum tolerance limits of mycotoxins in fresh citrus fruits and their derivatives. We are aware the latter is a more complex process and needs a multidisciplinary approach, as both the dietary habits of targeted populations and the additive effects of toxins from different food sources have to be taken into consideration. Regarding the risk of mycotoxin contamination of citrus fruits, an interesting finding is that with the analytical method we used a consistent level of secondary metabolites which were detected even in asymptomatic fruits. Moreover, the metabolic profile of the peel extract of asymptomatic fruits did not differ substantially from the profile of the peel extract of hail-damaged fruits. This is consistent with the results of isolations, which confirmed that Alternaria, Colletotrichum, and Penicillium reside in the peel of asymptomatic fruits as epiphytes, endophytes, or latent pathogens. UHPLC-Q-TOF-MS proved to be very effective as an analytical method for its rapidity, sensitivity, and robustness.
This study provided new insights into several theoretical and practical aspects concerning the biology and epidemiology of A. alternata, C. gloeosporioides, and P. digitatum and the identification of secondary metabolites that these fungi produce during the interaction with the host plant. In particular, this is the first study addressing the effects of hailstorms on fruit contamination by mycotoxins produced by fungi residing in the fruit peel as latent pathogens. In the light of the results of this study, the risk of contamination appears low for metabolites of A. alternata and C. gloeosporioides, even on fruits injured by the hail. Conversely, it seems real in fruits infected by P. digitatum as in both the peel and juice of decayed fruits we found considerable amounts of Penicillium-related toxins such as patulin and Rubratoxin B. Although mummified or severely blemished fruits usually do not enter the citrus fresh fruit supply chain as they are discarded before, this does not completely exclude the risk of contamination due to post-harvest rot. Moreover, mummified or severely blemished fruits are often present in bulk quantities destined for the juice industry. An accurate and severe selection of these fruits would prevent or reduce the risk of contamination by mycotoxins in the citrus juice industry.

Orange Fruit Sampling
On the 24 January 2022, mature fruits of two blood orange cultivars, 'Tarocco Lempso' and 'Tarocco Tapi', were collected in three 20-year-old commercial orchards cultivated with organic farming methods, in three diverse districts of the province of Catania (eastern Sicily). Around five months before the sampling, a hailstorm had affected the orchards, and since then trees had not been treated with fungicides. Three distinct types of fruits were selected per each cultivar: fruits injured by hail (fruits with necrotic lesions covering around 25% of the rind surface), mummified fruits still on the tree canopy, and asymptomatic fruits (fruits with no visible rind injury) ( Figure 12).
Overall, in each orchard 60 fruits of each orange cultivar were collected. Five fruits of each type (hail-injured, mummified, and asymptomatic) were picked up randomly from four randomly selected trees. Overall, 60 fruits of the same type and cultivar were pooled together, and three replicates, each of 20 fruits, for each composite sample were processed and analyzed separately (Table 4). On the 24th of January 2022, mature fruits of two blood orange cultivars, 'Tarocco Lempso' and 'Tarocco Tapi', were collected in three 20-year-old commercial orchards cultivated with organic farming methods, in three diverse districts of the province of Catania (eastern Sicily). Around five months before the sampling, a hailstorm had affected the orchards, and since then trees had not been treated with fungicides. Three distinct types of fruits were selected per each cultivar: fruits injured by hail (fruits with necrotic lesions covering around 25% of the rind surface), mummified fruits still on the tree canopy, and asymptomatic fruits (fruits with no visible rind injury) ( Figure 12). Overall, in each orchard 60 fruits of each orange cultivar were collected. Five fruits of each type (hail-injured, mummified, and asymptomatic) were picked up randomly from four randomly selected trees. Overall, 60 fruits of the same type and cultivar were pooled The sampling design is displayed in Table 4. Fruit samples were kept into plastic bags and immediately shipped to the Laboratory of Food Toxicology at the Department of Preventive, Medicine, Nutrition and Food Science Area of the University of Valencia, Spain. Samples were stored at 4 • C until isolation of fungi from the rind. Subsequently, the juice was extracted, and secondary metabolites from the peel and juice were analyzed separately.

Isolation of Fungi
Isolations were conducted from the fruit peel in accordance with the method of Riolo et al. [17], with few modifications. Fruits were washed with tap water. Their surface was sterilized with a sodium hypochlorite solution (10%) for 1 min. Fruites were immersed in 70% ethanol for 30 s and rinsed in sterile distilled water (s.d.w.). After disinfection, 15 peel fragments (5 × 5 mm) per fruit were taken with a scalpel (Figure 13). In fruits damaged by hail, fragments were taken from necrotic lesions of the rind, while in mummified and asymptomatic fruits, fragments were taken randomly from the entire surface of the fruit. The fragments were blotted dry with sterile filter paper and placed in Petri dishes (three Petri dishes per fruit) on Potato Dextrose Agar (PDA; Oxoid Ltd., Basingstoke, UK) amended with 150 µg/mL streptomycin and incubated for 2 days at 25 • C, in the dark. Pure cultures of each fungal isolates were obtained by single-conidium subcultures. fragments (5 × 5 mm) per fruit were taken with a scalpel (Figure 13). In fruits damaged by hail, fragments were taken from necrotic lesions of the rind, while in mummified and asymptomatic fruits, fragments were taken randomly from the entire surface of the fruit. The fragments were blotted dry with sterile filter paper and placed in Petri dishes (three Petri dishes per fruit) on Potato Dextrose Agar (PDA; Oxoid Ltd., Basingstoke, UK) amended with 150 µg/mL streptomycin and incubated for 2 days at 25 °C, in the dark. Pure cultures of each fungal isolates were obtained by single-conidium subcultures.

Morphological Characterization of Isolates
The macroscopic characteristics (color, margin, diameter, and texture) of colonies were examined according to the method of Pryor and Michailides [149], whereas microscopic features (conidium and conidiophore branching morphology) were examined according to the method of Simmons [150] and Agrios [151] (Figure 13). Macro-and microscopic identification were used to separate the isolates by genus.

Molecular Characterization of Isolates
Genomic DNA was extracted from single-conidium isolates using the procedure described by Riolo et al. (2021) [17]. Overall, 290 selected fungal isolates of three distinct genera obtained from orange fruits were examined. Isolates were identified by the amplification and multilocal analysis of specific barcode regions (see Table 5). The PCR amplifications were performed on a GeneAmp PCR System 9700 (Applied Biosystems, Monza-Brianza, Italy). All PCRs were performed by using the Taq DNA polymerase recombinant (Invitrogen™, Carlsbad, 254 CA, USA) and carried out in a total volume of 25 µL containing the following: PCR Buffer (1X), dNTP mix (0.2 mM), MgCl2 (1.5 mM), forward and reverse primers (0.5 µM each), Taq DNA Polymerase (1 U), and 1 µL of genomic DNA (1 ng/µL) [34]. Amplified products were analyzed by electrophoresis, and single bands of the expected size were purified with the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and sequenced with both forward and reverse primers by Colección Española de Cultivos tipo Parc Científic, Universitat de València (Valencia, España).

Preparation of Samples for the Analyses and Extraction of Orange Peel and Juice
Peel was manually separated from albedo using a knife, and the fresh orange juice was extracted from peeled and pressed oranges using a domestic squeezer. The peel and orange juice of each fruit were separated and individually transferred into conical polytetrafluoroethyl (PTFE) centrifuge tubes (15 mL). Before the extraction process, all the samples were stored in polyethylene tubes maintained at −80 • C.

Extraction of Fungal Secondary Metabolites from Orange Peel and Juice
The extraction of fungal secondary metabolites of orange peel was carried out in accordance with the method of Widiantini et al. [156], partially modified. At first, 5 mL of methanol per gram of orange peel was added into a conical polytetrafluoroethylene (PTFE) centrifuge tube (15 mL); then, it was shaken for 12 h at room temperature with an orbital laboratory shaker. The supernatant was collected and filtered through a 13-mm/0.22-µm nylon syringe filter (Membrane Solutions) in a 2 mL Amber Vial for chromatography analysis, and 20 µL of this solution was injected for UHPLC-Q-TOF-MS analysis.
The extraction of fungal secondary metabolites of orange juice was carried out according to the method of El Jai et al. [157], slightly modified. At first, 10 mL of orange juice and 20 mL of ethyl acetate (extraction solvent) were added in a conical polytetrafluoroethyl (PTFE) centrifuge tube (50 mL), and the suspension was mixed for 1 min; then, it was centrifuged for 10 min at 4000 rpm at 5 • C, using an Eppendorf centrifuge 5810R (Eppendorf, Hamburg, Germany). The organic phase at the top of the tube was recovered and placed in another PTFE centrifuge tube (15 mL) and saved. The total collected organic phase was evaporated to dryness under a nitrogen stream using a TurboVap LV evaporator (Zymark, Hopkinton, MA, USA). The dry extract was reconstituted with 1 mL of methanol and filtered through a 13-mm/0.22-µm nylon syringe filter (Membrane Solutions) in a 2 mL Amber Vial for chromatography analysis.

Analysis of Fungal Secondary Metabolites by UHPLC-Q-TOF-MS
Orange peel and orange juices extracts were injected for UHPLC-Q-TOF-MS analysis. All the analyses were performed in triplicate (n = 3). The samples were analyzed using an UHPLC device (1290 Infinity II LC, Agilent Technologies) composed of an automatic sampler, a binary pump, and a vacuum degasser, coupled with a quadrupole time of flight mass spectrometer (Agilent 6546 LC/Q-TOF), operating in positive ionization mode (Agilent Technologies, Santa Clara, CA, USA), which was used for chromatographic analysis. A total of 5 µL of sample was injected. Complete analysis was performed in 25 min. Chromatographic separation was performed with an Agilent Zorbax RRHD SB-C18, 2.1 mm × 50 mm, 1.8 µm column. Mobile phase A was composed of Milli-Q water, and acetonitrile was used for mobile phase B (both phases were acidified with 0.1% formic acid), with gradient elution, as follows: 0 min, 2% B; 22 min 95% B; and 25 min, 5% B. The column was equilibrated for 3 min before every analysis. The flow rate was 0.4 mL/min. Dual AJS ESI source conditions were as follows: gas temperature: 325 C; gas flow: 10 L/min; nebulizer pressure: 40 psig; sheath gas temperature: 295 C; sheath gas flow: 12 L/min; capillary voltage: 4000 V; nozzle voltage: 500 V; Fragmentor: 120 V; skimmer: 70 V; product ion scan range: 100-1500 Da; MS scan rate: 5 spectra/s; MS/MS scan rate: 3 spectra/s; maximum precursors per cycle: 2; and collision energy: 10, 20, and 40 eV. An untargeted LC/Q-TOF-based metabolomics approach was used to identify the different secondary metabolites of fungi growing on orange peel and orange juice. Integration, data elaboration, and identification of metabolites were managed using MassHunter Qualitative Analysis Software B.08.00 and library PCDL Manager B.08.00 [85].

Statistical Analysis
The analytical data obtained by UHPLC-Q-TOF-MS were log10 transformed before statistical analysis. Relationships among two different cultivars ('Tarocco Lempso' and 'Tarocco Tapi') × Alternaria, Colletotrichum, and Penicillium species combinations were analyzed using Pearson's correlation coefficient analysis. All the above statistical analyses were performed using RStudio v.1.2.5 (R). MetaboAnalyst 5.0 software [158] was used for principal component analysis (PCA) using log10 transformed data. The features included were log-transformed and mean-centered. Funding: This research was funded by the University of Catania, Italy "Investigation of phytopathological problems of the main Sicilian productive contexts and eco-sustainable defense strategies (ME-DIT-ECO) PiaCeRi-PIAno di inCEntivi per la Ricerca di Ateneo 2020-22 linea 2" "5A722192155", by the projects "Smart and innovative packaging, postharvest rot management, and shipping of organic citrus fruit (BiOrangePack)" under the Partnership for Research and Innovation in the Mediterranean Area (PRIMA)-H2020 (E69C20000130001), the "Italie-Tunisie Cooperation Program 2014-2020" project "PROMETEO «Un village transfrontalier pour protéger les cultures arboricoles méditerranéennes en partageant les connaissances»" cod. C-5-2.1-36, CUP 453E25F2100118000, and by the European Union (NextGeneration EU), through the MUR-PNRR project SAMOTHRACE (ECS00000022).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable. Data Availability Statement: Data will be available upon specific request to the authors.