Characterization of Citrus-Associated Alternaria Species in Mediterranean Areas

Alternaria brown spot is one of the most important diseases of tangerines and their hybrids worldwide. Recently, outbreaks in Mediterranean areas related to susceptible cultivars, refocused attention on the disease. Twenty representatives were selected from a collection of 180 isolates of Alternaria spp. from citrus leaves and fruit. They were characterized along with reference strains of Alternaria spp. Micro- and macroscopic characteristics separated most Alternaria isolates into six morphotypes referable to A. alternata (5) and A. arborescens (1). Phylogenetic analyses, based on endopolygalacturonase (endopg) and internal transcribed spacer (ITS), confirmed this finding. Moreover, a five-gene phylogeny including two anonymous genomics regions (OPA 1–3 and OPA 2–1), and the beta-tubulin gene (ß-tub), produced a further clustering of A. alternata into three clades. This analysis suggested the existence of intra-species molecular variability. Investigated isolates showed different levels of virulence on leaves and fruit. In particular, the pathogenicity on fruit seemed to be correlated with the tissue of isolation and the clade. The toxigenic behavior of Alternaria isolates was also investigated, with tenuazonic acid (TeA) being the most abundant mycotoxin (0.2–20 mg/L). Isolates also synthesized the mycotoxins alternariol (AOH), its derivate alternariol monomethyl ether (AME), and altenuene (ALT), although to a lesser extent. AME production significantly varied among the six morphotypes. The expression of pksJ/pksH, biosynthetic genes of AOH/AME, was not correlated with actual toxin production, but it was significantly different between the two genotypes and among the four clades. Finally, ten isolates proved to express the biosynthetic genes of ACTT1 phytotoxin, and thus to be included in the Alternaria pathotype tangerine. A significant correlation between pathogenicity on leaves and ACTT1 gene expression was recorded. The latter was significantly dependent on geographical origin. The widespread occurrence of Alternaria spp. on citrus fruit and their ability to produce mycotoxins might represent a serious concern for producers and consumers.


Introduction
The fungal genus Alternaria is widely and abundantly disseminated in the environment (atmosphere, soil, seeds, and agricultural produce) [1]. It comprises numerous saprophytic, endophytic, and pathogenic species, causing pre-and postharvest deterioration of cereals, fruit, and vegetables. For example, Alternaria brown spot is a highly destructive disease of tangerines and tangerine hybrids of worldwide importance [2]. The disease is prevalent in citrus production areas with a Mediterranean climate, characterized by cool, humid winters and hot, arid summers [3]. In Europe, it has been reported in Spain [4], Italy [5], and Greece [6]. It attacks young leaves, twigs and fruit, causing brown to black lesions surrounded by a yellow halo. Severely infected leaves and fruit may fall, and whole shoots may wilt and die. Therefore, under appropriate environmental conditions, significant losses might occur in terms of both yield and marketability of fruit of susceptible cultivars [2,7]. Alternaria spp. can also cause citrus black rot, which occurs mostly as a core rot [8], as well as leaf spot on rough lemon, and Mancha foliar on Mexican lime. However, these two latter diseases are considered of low economic significance because they are limited to nursery and seed production blocks, or occur in very limited areas [3]. Both black rot and brown spot are caused by several 'small-spored' Alternaria species [7,9]. Recently, 35 of them were synonymised as Alternaria alternata, meaning that Alternaria sect. Alternaria consists of only 11 species (A. burnsii, A. tomato, A. jacinthicola, A. iridiaustralis, A. eichhorniae, A. betae-kenyensis, A. gaisen, A. alstroemeriae, A. longipes, A. gossypina, and A. alternata) and one species complex (A. arborescens) [10]. Moreover, the use of formae specialis (f. sp.) to indicate isolates morphologically indistinguishable from A. alternata but infecting specific hosts has been suggested. In particular, at least 16 different f. sp. epithets occur in literature, e.g. A. alternata f. sp. mali, or A. alternata f. sp. citri.
For classification purposes, the production of the numerous secondary metabolites is also used. They are not only significant in terms of food quality and safety, but also play an important role in disease occurrence [1,11,12]. For instance, the host-selective toxins (HSTs), lowmolecular-weight compounds with a wide spectrum of structures, determine host range or specificity, since they are toxic only to specific plant species, varieties, or genotypes [13]. As these toxins are required to invade tissues and cause disease, they represent pathogenicity factors [14,15]. The tangerine pathotype of A. alternata, causing brown spot, produces HSTs known as ACT-toxins. The structure of ACT-toxins is closely related to those of AK-and AFtoxins, which are HSTs produced by the Japanese pear and strawberry pathotypes of A. alternata, respectively. Two genes controlling AK-toxin biosynthesis (AKT1 and AKT2) have been cloned [16], and more recently, their two ACT-toxin homologues (ACTT1 and ACTT2), involved in the biosynthesis of the decatrienoic acid moiety [16], have been sequenced [17].
Among secondary metabolites, there are also several mycotoxins, which might pose a risk to human health [11]. The ones occurring most frequently in nature are alternariol (AOH) and its derivate alternariol monomethyl ether (AME), tenuazonic acid (TeA), tentoxin (TEN), altenuene (ALT) and altertoxin I, whereas iso-altenuene and altertoxin II have not been found in crops to date. TeA has cytotoxic and phytotoxic properties, and its LD 50 value in mice is similar to that of Fusarium deoxynivalenol [18]. Altertoxins (ALXs) are considered extremely toxic and stronger mutagens to mice than AOH and AME [19]; moreover, the high genotoxicity of ALX II to mammalian cells has been proven [20,21]. AOH is formed along the polyketide route, which is a common pathway for the formation of many secondary fungal metabolites [22,23]. Biosynthetic routes for AOH have been studied extensively [24,25]. More recently, in a draft genome sequence of A. alternata, 10 putative polyketide synthases (pks)-encoding genes have been identified [22]. Among them, pksJ proved to be required for biosynthesis, whereas, pksH downregulation proved to affect pksJ expression, thus having an indirect effect on AOH production [22].
In the present study, specific investigations were made to gain insight into recent Alternaria disease outbreaks in the Mediterranean basin. The aim of the study was to characterise the main species involved, also in relation to their ability to produce mycotoxins, in order to draw up appropriate control strategies.

Ethics Statement
This study did not involve endangered or protected species. No ethical permits were required for this work, which involved no experimentation on animals or human samples.

Isolates used in the study
Twenty representative isolates were selected from the Alternaria spp. collection of the Department of Soil, Plant and Food Sciences, University of Bari Aldo Moro, Italy. In particular, they were selected among citrus isolates and according to tissue (leaf/fruit) and geographical area of isolation, i.e. Italy (GPS coordinates 40°20'39.2"N and 16°43'18.5"E) and Spain (GPS coordinates 39°56'42.993"N and 0°3 0 38.389@ W) ( Table 1).

Morphological characterization
Selected isolates were characterized along with reference strains of A. alternata ( To determine colony morphology, plates were incubated for 10 days at 22±1°C in the dark. The macroscopic characteristics (colour, margin, diameter, and texture) were analysed as reported by Pryor and Michailides [26].
To study the microscopic characteristics, after 5 days of incubation in the dark, a rectangular block (10×20 mm) of agar and mycelium was removed aseptically within the expanding margin of the colony, exposed to daylight and returned to incubation surface. On day 10, it was put on a slide and gently pressed with a glass slip. For each isolate, observations were made at ×40 magnification and sub-stage illumination to highlight the sporulation apparatus features. The sporulation characteristics were compared with those reported in the Alternaria identification manual [27] and by Pryor and Michailides [26].

Molecular characterization
A multilocus approach was adopted to characterize and determine the phylogenetic collocation of the 20 selected isolates from citrus. Portions of five Alternaria barcoding genes/regions, i.e. internal transcribed spacer (ITS), endopolygalacturonase (endopg), two SCAR markers (OPA 1-3 and OPA 2-1), and beta-tubulin (ß-tub), were sequenced. Fungal isolates were grown for 7 days on PDA at 24±1°C in the dark, and mycelia were collected and stored at -20°C until use. Genomic DNA was extracted from 100 mg of mycelium as reported by Sanzani et al. [28], and diluted to 50 ng/μl in sterile ultra-pure water. PCR was carried out in a 50 μl reaction mixture containing 50 ng of DNA, 10 μM of each primer, and 5U of Taq DNA Polymerase (Emeral-dAmp PCR master mix, Takara, Clontech, USA). PCR was carried out in a thermal cycler (MyCycler, BioRad, Hercules, CA, USA), using reported cycling conditions [8,29,30,31]. All primer pairs were synthesized by Sigma-Aldrich (Milan, Italy).
Amplicons were resolved in 1% agarose gels in TAE buffer (1×) and visualized by UV illumination. DNA was recovered from agarose by Isolate II Genomic DNA kit (Bioline, London, UK) according to manufacturer recommendations. Purified products were sequenced with both forward and reverse primers by Macrogen Europe (Amsterdam, The Netherlands). CHROMASPRO v. 1.5 software (http://www.technelysium.com.au/) was used to evaluate sequence reliability, and the consensus sequences were deposited in Genbank (Table 1). Unreliable sequences containing doubtful bases were re-sequenced.
For the molecular identification of isolates, ITS and endopg sequences obtained in the present study, and validated sequences representative of all species identified within Alternaria Sect. Alternaria were phylogenetically analysed to determine the taxonomic status of isolates in the light of recent molecular criteria for classification [10]. Before analyses, the complete panel of reference sequences was analysed utilizing the ElimDupes software (http://hcv.lanl.gov/ content/sequence/ELIMDUPES/elimdupes.html) to delete multiple identical sequences. Identical reference sequences were included in the panel when representative of different Alternaria species [10]. Sequences were aligned using MUSCLE and introduced to MEGA6 for phylogenetic analysis with the Maximum Likelihood method using the Tamura-Nei model [32]. Analyses were performed with 500 bootstrap replications.
In order to maximize the effectiveness of the investigation into the genetic diversity within isolates from citrus obtained in the present study, the phylogenetic analysis was repeated using a combined data set of all sequenced markers (ITS, endopg, OPA 1-3, OPA 2-1, and ß-tub).

Pathogenicity tests
The ability of isolates to cause brown spot was assessed in vivo on freshly harvested citrus fruit and leaves (Citrus reticulata cv. Fortune), free of defects or injuries, and uniform in size, colour and ripeness, collected in Metaponto (Italy, GPS coordinates 40°23'17.611''N, 16°47'48.799''E).
The selected citrus fruit and leaves were randomized, surface-sterilized with 1% sodium hypochlorite, and extensively washed under running tap water. After air-drying, they were wounded and kept at room temperature for 30 min; each fruit was wounded with a sterile nail (3×3 mm) in two equidistant points on the equatorial surface, leaves were wounded by pricking the abaxial side with 3 entomological pins mounted on a cork. For each isolate, a drop (10 μl) of conidial suspension (10 5 conidia/ml), prepared by flooding plates with 5 ml sterile 0.02% Tween20 solution and gently rubbing the colony surface with a sterile spatula, was applied into the wounds of 24 fruit and 18 leaves that were then randomly divided into three replicates. Samples were incubated at 24±1°C and high relative humidity for 10 days. The disease incidence (infected wounds, %) and severity (lesion diameter, mm) were recorded. The assay was repeated twice.

Expression analyses
The selected isolates (in triplicate) were grown in 30 ml of Potato Dextrose Broth (PDB, Oxoid) on a rotary shaker (150 rpm), in the dark at 24±1°C. After 10 days, the mycelium was harvested, lyophilized and kept at -80°C. The culture broth was stored at -20°C until mycotoxin analysis.
Total RNA was extracted from 100 mg of mycelium using Plant RNA/DNA Purification Kit (Norgen Biotek, ON, Canada) and purified by treatment with DNaseI following manufacturer's instructions. The total RNA obtained was visualized on a 1.5% (w/v) agarose gel and quantified by a NanoDrop Spectrophotometer (Thermo Scientific, Wilmington, USA). Then, it was serially diluted in the range 0.1-1000 ng, and the dilutions were reverse-transcribed using an iScript cDNA synthesis kit (BioRad). Primers obtained from regions of genes ACTT1 and ACTT2 (phytotoxins), and pksJ and pksH (mycotoxins) were used in real time quantitative PCR (qPCR) reactions (Table 2).
Standard curves were generated by plotting Cycle threshold (Ct) values (y-axis) against logs of total RNA (x-axis). Reaction efficiency, precision, and dynamic range were calculated [33]. The cDNA used for the expression reactions was synthesised from 500 ng total RNA. The amplification mixture (10 μl) contained 1 μl of template, 5 μl of SensiFAST SYBR & Fluorescein Mix (Bioline), 3.2 μl of ultra-pure H 2 O and 0.4 μl of each specific primer (10 μM). In negative control samples, cDNA was replaced with sterile water or non-reverse transcribed RNA to exclude possible cross-contamination and the presence of genomic DNA, respectively. Amplifications were performed in 96-well reaction plates using an iCycler iQ thermal cycler (BioRad). The amplification program consisted of 1 cycle of 95°C for 2 min, followed by 40 cycles of 95°C for 5 s, 62 (pksJ-pksH)/60 (ACTT1-ACTT2)°C for 10 s and 72°C for 10 s. Subsequently, a melting curve analysis was done to determine the reaction specificity by incubating the samples at 95°C for 2 min, annealing at 65°C for 5 s, followed by heating them slowly at 0.5°C/sec up to 95°C, while continuously monitoring the fluorescence signal. Moreover, an aliquot (5 μl) of product was run on a 1% agarose gel. Ct was automatically generated by the iCycler associate software (Real Time Detection System Software, version 3.0). The absolute expression was calculated by interpolation from the calibration curve.

Mycotoxin analyses
A method for the simultaneous detection of the main five Alternaria toxins (ATs) was applied [34]. A liquid chromatography-triple quadrupole mass spectrometer (Applied Biosystems, Foster City, CA, USA) equipped with atmospheric pressure chemical ionisation (APCI) was used. The culture broth was filtered through filter paper (Whatman No 4, GE Healthcare, Milan, Italy) and re-filtered using a sterile 45μ syringe filter (Bibby Scientific Limited, Stone, Staffordshire, UK). Filtrates were extracted as reported by Prelle et al. [34] and analysed as mentioned above. The entire experiment was repeated twice.

Statistical analysis
All statistical analysis were conducted in R (https://www.r-project.org/). Ten data samples in replicates, i.e. pathogenicity on leaf, pathogenicity on fruit, four toxins, and pksH, pksJ, ACTT1 and ACTT2 expression, were independently tested for normality using the Shapiro-Wilk (SW) test and Q-Q plot inspection. Given that samples did not come from normally distributed populations (SW test p < 0.05), nonparametric tests were chosen for downstream analyses. Each sample was divided into distinct classes, i.e. two origins, two tissues, six morphotypes, two genotypes and four clades. The two-ways Wilcoxon (W) test was applied for comparing samples from two classes and the Kruskal-Wallis (KW) test for three or more classes. A Heatmap was built on the full data matrix using default parameters in ClustVis (http://biit.cs.ut.ee/clustvis/).

Morphological characterization of isolates
All 20 isolates were grouped according to macro-and microscopic features on PDA, MEA, and PCA, along with reference strains of A. alternata and A. arborescens from CBS-KNAW. A higher variability in colony morphology was observed on PDA, as compared to PCA and MEA (data not shown), allowing the differentiation of six morphotypes. In particular, twelve isolates (A22, A23, A25, A27, A28, A29, A31, A45, A63, A64, A65, A67) showed colonies that were flat, woolly and with colours ranging from brown to black, and an average diameter of 65 mm. They produced dark brown conidia arranged in branched chains. Conidia appeared oval-ellipsoidal with 3-5 transverse septa. These features matched those of A. alternata reference strain CBS 112249 (morphotype 1).
Two isolates (A24, A43) showed colonies varying from greenish-grey to brown, characterized by a lower growth rate (average diameter 45 mm after 7 days on PDA). Conidia appeared oval or ellipsoidal with 1-4 transverse septa and 1-2 longitudinal septa. Conidia were borne by long primary conidiophores, occasionally presenting sub-terminal branches. These characteristics matched those of the reference strain for A. arborescens CBS 109730 (morphotype 2).

Molecular characterization
A two-gene phylogeny, including ITS and endopg sequences, of isolates selected in the present study and reference CBS strains [10] was performed to identify isolates (Fig 1). According to this analysis, 18 out of 20 isolates were associated to A. alternata. They were almost equally distributed as regards the two geographical areas (Spain or Italy) and tissue of isolation (leaf or fruit), and clustered with several reference isolates of A. alternata, including the strain CBS 112252 (Fig 1). Finally, two isolates namely A24 (Spain, fruit) and A43 (Italy, leaf) grouped within the species complex of A. arborescens along with strains CBS 105.24 from Solanum tuberosum and CBS 112749 from Malus domestica.
The phylogenetic analysis of isolates using all sequenced genes (ITS, endopg, ß-tub, OPA1-3 and OPA2-1) made it possible to achieve a more accurate discrimination and to identify four main clades (Fig 2). In particular, three clades, represented by 13, 3 and 2 isolates, respectively, were identified in A. alternata (Fig 2). One additional clade, represented by isolates A24 and A43, corresponded to the A. arborescens species complex. A partial matching was found between molecular clades and morphotypes. In particular, isolates of clade II belonged to morphotype 1 (alternata), which was also the prevalent morphotype in clade I. The latter included also morphotypes 3 and 5 (ex-tenuissima and ex-citri). Isolates of the A. arborescens species complex (clade IV) were characterized by the same morphotype (2). Finally, clade III included morphotypes 4 (ex-limoniasperae) and 6 (ex-toxicogenica).

Pathogenicity Tests
Pathogenicity assays foresaw the evaluation of disease incidence and severity up to 10 days from inoculation on both fruit and leaves. All tested isolates caused necrosis around the inoculation point, expanding into leaf and fruit tissues (Fig 3). Similar necrotic symptoms were observed, regardless of origin and genetic groupings. The lesions developed initially as rapidly enlarging, almost circular, brown, water-soaked spots that quickly turned into dark brown, necrotic spots surrounded by yellowish margins. Lesions on leaves, ranging from 17 to 42 mm, were larger than those on fruit, ranging from 5 to 18 mm (W test p < 0.001). The pathogenicity on fruit was slightly higher in leafderiving isolates compared with fruit-deriving isolates (W test p < 0.05), and it also differed among the isolates from the four clades (KW test p < 0.005); instead, the pathogenicity on leaves was not found to differ in terms of geographical origin, tissue of isolation, morphotype, genotype or clade.

Phytotoxin and mycotoxin biosynthetic gene expression
The expression analysis of the genes ACTT1 and ACTT2 is reported in Fig 4A. Primer pairs were designed upon gene sequences available in Genbank and for each one standard curves were drawn up ( Table 2). An absolute quantification of gene expression was performed by interpolation from the standard curves. Specific qPCR reactions were conducted for the 20 selected isolates. After 10 days of incubation at 24°C, ACTT1 was expressed at detectable levels only in 10 isolates (i.e. A22, A23, A24, A26 A28, A29, A30, A31, A41, A45) and the absolute quantity of total RNA was 1-4 log (ng). Similarly, ACTT2 was expressed in 4 out of the 20 tested isolates, namely A22, A28, A29 and A30, and the absolute quantity of total RNA was 1-6 log (ng). PCR reactions were conducted to confirm the absence of coding genes (data not shown).
Moreover, specific qPCR reactions were performed using primer pairs for pksJ and pksH, whose standard curves were drawn up for absolute quantification ( Table 2). Results showed that, after 10 days of incubation at 24±1°C, there was a 1-3.5 log of transcript (ng) for pksH, and a 1.5-3.5 log for pksJ (Fig 4B).
The expression of the two pksJ and pksH genes was linearly correlated (linear regression R 2 = 0.78, p < 0.0001), and A43 (A. arborescens) was the isolate in which the expression of the two genes was lower. The expression of each gene was higher in Italian isolates compared to Spanish isolates (W test p < 0.05), and it significantly varied among different clades (KW test p < 0.0005).

Mycotoxin analyses
The toxigenic potential of the selected Alternaria spp. isolates was investigated in vitro on liquid cultures (Table 3). In particular, they were assayed for the production of TeA, AOH, AME, ALT, and TEN. All tested isolates proved to produce TeA, AOH, AME, and ALT, whereas TEN was not detected in any of the samples. Overall, TeA was the most abundant toxin, produced in the range of 0.2-20 mg/l, with A23 (A. alternata) and A43 (A. arborescens) being the highest and lowest producers, respectively. TeA was 1.6-fold more abundant in isolates from fruit compared to isolates from leaves (W test p < 0.05). A consistent production of ALT was also detected, in the range of 0.08-8 mg/l. ALT levels were 3-fold higher in Italian isolates compared with Spanish ones (W test p < 0.005), 2-fold higher in fruit isolates than in leaf ones (W test p < 0.001), and also varied among distinct clades (KW test p < 0.05). AOH and its derivate product AME were detected in all samples in the range of 30-200 and 20-300 μg/l, respectively. The highest AOH producers were A23 and A44; AME production differed significantly among distinct morphotypes (KW test p < 0.005) and clades (KW test p < 0.05).
The heatmap analysis clustered the full data set according to isolates and variables (Fig 5). The cluster of variables grouped ACTT gene expression with pathogenicity on leaves, whereas the expression of both pks genes grouped with pathogenicity on fruit and AME. The other mycotoxins constituted an out-group with a poorly uniform pattern of values.

Discussion
This work encompasses studies on the diversity of Alternaria isolates recovered from citrus plants (fruit and leaves) affected by brown spot in Mediterranean countries relevant to citrus production. They were conducted using morphological, pathogenic, toxigenic, and molecular approaches, since only a polyphasic approach could be useful in the characterization of the genus Alternaria [35,36]. Concerning morphological characterization, interesting results were observed when isolates were grown on PDA. Indeed, as also reported by Pryor and Michailides [26], on PDA Alternaria produces conidiophores that are much less dense as compared to other cultural media, and thus a better observation of typical structures and associated conidial chains is possible. Moreover, the use of ready-to-use commercially available media guarantees greater batch-to-batch uniformity as compared to locally prepared non-commercial formulations, although the firmness and reproducibility of morphological descriptions would surely benefit greatly from the development of a specific medium suitable for the sporulation of Alternaria spp. [26]. Although we are aware of studies concerning the influence of light/dark cycle on toxin production [37], they refer to another growth medium (mCDB) than the one used in the present investigation (PDA/PDB). Pruß and co-workers [37] found that AOH and AME were produced in considerable amounts even in the dark. Moreover, in other similar papers, Alternaria strains were incubated in the dark prior to mycotoxin analysis [38,39,40]. Since the growth conditions of our isolates were the same in all assays, morphology, pathogenicity, and toxigenicity results could be compared.
Almost all isolates included in the study proved to match with A. alternata, which is the most widespread species across plants, seasons, and geographical regions, including host-specific pathogenic strains, as well as opportunistic and saprophytic forms causing the spoilage of freshly harvested crops [41]. Within this species, in the present study five morphotypes were identified according to colony and sporulation apparatus features, which correlated well with the CBS reference strains included for comparative purposes, although there were minor differences among isolates within each morphotype. In particular, morphotype 1 (A. alternata) was more abundant than the others, independently from geographical or tissue origin. However, a higher number of morphotypes was recorded among Italian isolates.
On the basis of the work by Woundenberg et al. [10], phylogenetic analyses were initially conducted by combining ITS region and endopg gene. It made it possible to cluster 18 out of 20 isolates with A. alternata CBS references, and the remaining two isolates with A. arborescens CBS references, thus confirming that isolates of the arborescens species complex are molecularly distinct from isolates of the alternata species-group. However, in the present study we observed that when further genes/regions (β-tub, OPA 1-3, OPA 2-1) were included in the analysis, a higher variability within A. alternata species was found. In particular, three clades were obtained, which, however, did not perfectly overlap with the morphotype grouping. Generally, the highest variability was observed in OPA1-3 region sequences, as already reported by Stewart et al. [42]. These results should be taken into account in the future studies on the characterization of Alternaria.
All of the investigated isolates were pathogenic to both leaves and fruit, although isolates proved to be more virulent on leaves. On the contrary, Huang et al. [41] and Peever et al. [7] found 15 and 8%, respectively, of non-pathogenic strains under inoculum conditions in their collections. A relation was found between pathogenicity on fruit and clades, with clade I the most pathogenic. Moreover, leaf isolates were slightly more pathogenic on fruit than the other isolates. This finding is not surprising, since leaf isolates expressed more phytotoxin genes that are known to be involved in Alternaria disease outbreak and progression. In particular, the tangerine pathotype produces ACTT1 and ACTT2, with toxin 1 being the most abundant [43]. A correlation was found between pathogenicity on leaves and ACTT1 expression, which was to be expected since ACTT1 is toxic to both citrus and pear (causing venal necrosis and a rapid increase in electrolyte loss from susceptible leaves), whereas ACTT2, the 5@-deoxy derivative of ACTT1, is highly toxic to pear and slightly toxic to citrus [44]. In our study, 10 isolates (eight Spanish and two Italian) expressed ACTT1 gene and four of them (A22, A28, A29, A30, all Spanish) also expressed ACTT2 gene. Morphotype 1 was the most represented. We are attempting to quantify the actual phytotoxin presence in inoculated tissues; however, the analysis is made difficult on account of the lack of a true standard.
Despite the well-known role of ACTs as pathogenicity factors, all of the tested isolates were able to cause brown spot, even in the absence of ACTT1/ACTT2 expression. These isolates proved to produce mycotoxins (TeA, ALT, AOH, and AME), whose role in several host-pathogen interactions has been analysed. They have been reported to be involved in the disease appearance/progression in Penicillium expansum-apple [23], Fusarium graminearum-cereals [45], but also in A. alternata-tomato [46] interactions. Therefore, mycotoxins might have played a role in the pathogenicity of the selected isolates. Indeed, as evidenced also by heatmap analysis, the expression of pksJ and pksH involved in AOH/AME biosynthesis was related to pathogenicity on fruit. Moreover, pks expression in Italian isolates (only two expressing ACT genes) was higher than in Spanish ones. Our feeling is that mycotoxins effect could sum up to phytotoxin in disease outcome, meaning that in the absence of phytotoxins, a certain degree of virulence is in any case guaranteed. However, this hypothesis needs to be validated by specific assays. In addition, the pks expression seemed to be related to the clade.
Concerning mycotoxin production, TeA was the most abundant one, as already reported for tomatoes by Van de Perre et al. [47]. An inter-clade variability was present for AME, as well as a difference among morphotypes for both AOH and AME. ALT was more abundant in Italian isolates coming from fruit. However, as also reported by Andersen et al. [48], chemotaxonomy alone cannot be considered a reliable tool for Alternaria identification. Considering the ability of the isolates to produce mycotoxins in considerable amounts, and the increasing concern of consumers and distributors as to their presence in both fresh and processed products, in the near future there will be a need for regulatory limits and monitoring strategies not only for crops, including citrus, but also processed food products (juices, purees, etc.). In fact, a dietary exposure assessment, performed for TeA with Belgian consumption data, revealed a mean value (4230 ng/kg bw/day) higher than the threshold value of toxicological concern (TTC) of 1500 ng/kg bw/day set by EFSA [11].

Conclusions
In conclusion, Alternaria-citrus represents an interesting pathosystem, since the fungus is a common saprophyte on citrus leaves and fruit in the grove, but it is able to turn into a pathogen under suitable conditions. In our study, most of the isolates belonged to A. alternata f. sp. citri pathotype tangerine morphotype alternata (S1 Table). Overall, a variability in morphotypes was observed among isolates from Italy. Pathogenicity was stronger on leaves than on fruit, and appeared to be related not only to ACTs, more abundantly produced on leaves, but also to other pathogenicity factors such as mycotoxins, produced by all isolates. This finding seems particularly true as far as pathogenicity on fruit is concerned. The different parameters analysed were not often related. It must be remembered that mycotoxin production depends on the geographical location where the plant is grown and harvested as well as on climate, the type of plant, and cultivar affected [40]. Similarly, Perrone et al. [49] found that a great variability was present in the parameters investigated (mycotoxin contamination, altitude, distance from the sea, relative humidity, etc.) among all evaluated grape-producing regions and sometimes within the same region.
In an effort to counter citrus diseases, the development of resistant cultivars through traditional breeding or genetic manipulation appears feasible, since the resistance of citrus to Alternaria brown spot is inherited as a recessive trait. Moreover, in the short-term further investigations into the epidemiology, fungicide activity, and application timing might improve disease control.
Supporting Information S1