In Vitro Fumonisin Biosynthesis and Genetic Structure of Fusarium verticillioides Strains from Five Mediterranean Countries

Investigating the in vitro fumonisin biosynthesis and the genetic structure of Fusarium verticillioides populations can provide important insights into the relationships between strains originating from various world regions. In this study, 90 F. verticillioides strains isolated from maize in five Mediterranean countries (Italy, Spain, Tunisia, Egypt and Iran) were analyzed to investigate their ability to in vitro biosynthesize fumonisin B1, fumonisin B2 and fumonisin B3 and to characterize their genetic profile. In general, 80% of the analyzed strains were able to biosynthesize fumonisins (range 0.03–69.84 μg/g). Populations from Italy, Spain, Tunisia and Iran showed a similar percentage of fumonisin producing strains (>90%); conversely, the Egyptian population showed a lower level of producing strains (46%). Significant differences in fumonisin biosynthesis were detected among strains isolated in the same country and among strains isolated from different countries. A portion of the divergent FUM1 gene and of intergenic regions FUM6-FUM7 and FUM7-FUM8 were sequenced to evaluate strain diversity among populations. A high level of genetic uniformity inside the populations analyzed was detected. Apparently, neither geographical origin nor fumonisin production ability were correlated to the genetic diversity of the strain set. However, four strains from Egypt differed from the remaining strains.


Introduction
Fusarium verticillioides (Sacc.) Nirenberg is a member of the Gibberella fujikuroi species complex, also called Fusarium fujikuroi species complex (FFSC), a group of 40 closely related Fusarium species defined by morphological traits, sexual compatibility and DNA-based phylogenetic analysis [1,2].
In particular, F. verticillioides belongs to the "African" clade of the FFSC [3], and it is the main causal agent of Fusarium ear rot of maize (Zea mays L.) [4,5]. This fungus has been reported worldwide and, in particular, it prevails in drier and warmer climatic regions [6,7] such as those present in temperate, (i) investigate the abilities of selected F. verticillioides strains isolated from maize kernels in five Mediterranean countries to in vitro biosynthesize FB 1 , FB 2 and FB 3 ; (ii) characterize the genetic structure of these selected strains to assess for possible variability within strains originating from each of the surveyed countries and between the strains originating from different countries.

Fungal Strains
A total of 90 F. verticillioides strains (Table 1) isolated from single maize kernels harvested from different fields in five Mediterranean countries (22 from Italy, 9 from Spain, 16 from Tunisia, 28 from Egypt and 15 from Iran) were used in this study (Figure 1). Isolation operations were carried out in the country of origin where all strains were properly stored in fungal collections. The investigated strains had not been extensively subcultured, thus avoiding possible alterations in fumonisin production. Some of the Italian strains used in this work had been already investigated in a previous study [50] and were included to further characterize them in a wider geographical context ( Figure 1). Table 1. Strain ID, country of origin and fumonisin B 1 , fumonisin B 2 and fumonisin B 3 production (µg/g) with standard errors (±SE) by Fusarium verticillioides strains isolated from maize kernels harvested in five Mediterranean countries and analyzed in this study.

Confirmation of F. verticillioides Identity by PCR Assays
To preliminarily confirm the identity of the 90 F. verticillioides strains used in this study, species-specific PCR assays were conducted. All strains were grown on Potato Dextrose Agar (PDA (Biolife Italiana, Milan, Italy)) at 22 • C for 14 d in the dark. DNA was extracted as described by Beccari et al. [56,57]. PCR assays were carried out with the specific primers VERT1 (GTCAGAATCCATGCCAGAACG) and VERT2 (CACCCGCAGCAATCCATCAG) [58]. A single PCR protocol was optimized using a total reaction of 20 µL. Each reaction contained 9.2 µL of sterile water for molecular biology (5prime, Hilden, Germany), 1.5 µL of cresol red (Sigma-Aldrich, Saint Louis, MO, USA), 2 µL of 10X PCR buffer (Microtech, Pozzuoli, Naples, Italy), 1.2 µL of magnesium chloride (Microtech), 2 µL of 10 mM DNTP mix (Microtech), 1 µL of 10 µM forward and reverse primers, 0.1 µL of 5 U/µL Taq polymerase (Microtech) and 2 µL of template DNA. The PCR cycle consisted of an initial denaturation step at 94 • C for 2 min, followed by 30 cycles of denaturation (94 • C for 35 s), annealing (60 • C for 30 s), extension (72 • C for 2 min) and a final extension at 72 • C for 5 min. PCR assays contained a positive control (template DNA of F. verticillioides) and a negative control with no DNA added. The amplification was performed in a T-100 thermal cycler (Bio Rad, Hercules, CA, USA). All PCR fragments were separated by electrophoresis by applying a tension of 110 V for about 1 h. Electrophoretic runs were visualized using an UV Image analyzer (Euroclone, Pero, Milan, Italy).

F. verticillioides Cultures
To determine in vitro fumonisin biosynthesis, cultures of F. verticillioides strains were obtained following the protocol indicated by Covarelli et al. [50] with slight modifications. In brief, 15 g of finely ground maize grains and 15 mL of deionized sterile water were added into 100 mL glass flasks (Duran, Mainz, Germany) to obtain the right moisture for allowing fungal development and then autoclaved three times at alternate days. Three flasks (replicates) for each F. verticillioides strain were then inoculated with a mycelium plug (0.6 cm diameter) taken from the growing edge of one-week-old pure fungal cultures of each strain developed on PDA at 22 • C in the dark. Three flasks (replicates) were used as controls by adding only a PDA plug. Flasks were incubated in the dark at 22 • C for 4 w, and developed cultures were then freeze-dried for 24 h using a lyophilizer instrument (Heto Powder Dry LL3000, Thermo Fisher Scientific, Waltham, MA, USA), ground with mortar and pestle and stored at −80 • C until further analysis.

Fumonisin Extraction and LC-MS/MS Analysis
Each fungal culture was extracted and analyzed in triplicate according to the validated and routine procedure also described by Covarelli et al. [50] with slight modifications. In brief, 1 g of ground sample was extracted with 5 mL of methanol/water (75:25, v/v) following 60 min shaking. The extract was filtered through filter paper. Prior to liquid chromatography, tandem Mass Spectrometry (LC-MS/MS) analysis, the extract was diluted by default 1:50 with a mixture of methanol/water (60:40), then filtered through 0.45 µm syringe filter. Twenty microliters were injected into the LC-MS/MS apparatus. If fumonisin levels were out of the calibration range, a further dilution (1:500 or 1:5000) was applied to the raw extract and then re-analyzed.
LC-MS/MS analyses were performed on a QTrap MS/MS system, from Applied Biosystems (Foster City, CA, USA), equipped with an Electrospray Ionization (ESI) interface and a 1100 series micro-Liquid Chromatography system comprising a binary pump and a micro-autosampler from Agilent Technologies (Waldbronn, Germany). The analytical column was a Gemini ® C18 column (150 × 2 mm, 5 µm particles) (Phenomenex, Torrance, CA, USA), preceded by a Gemini ® C18 guard column (4 × 2 mm, 5 µm particles). The column oven was set at 40 • C. The flow rate of the mobile phase was 200 µL/min, and the injection volume was 20 µL.
The column effluent was directly transferred into the ESI interface, without splitting. Eluent A was water and eluent B was methanol, both containing 0.5% acetic acid. A gradient elution was performed as follows. The percentage of eluent B was increased from 40% to 80% in 10 min, kept constant 3 min, then increased to 100% in 1 min, and kept constant for 4 min. The column was re-equilibrated with 40% eluent B for 7 min. The ESI interface was used in positive ion mode with the following settings: temperature 350 • C; curtain gas, nitrogen, 30 psi; nebulizer gas, air, 10 psi; heater gas, air, 30 psi; ion spray voltage +4500 V. The mass spectrometer operated in Multiple Reaction Monitoring (MRM) mode. Mycotoxin quantification was performed by external calibration in neat solvent. The identity of fumonisins was confirmed by comparison with the analytical standard considering chromatography retention time and MRM transitions (ion ratios) in agreement with the official guidelines for mycotoxin identification by Mass Spectrometry [59]. Detection limits in maize fungal cultures were 0.002 µg/g for FB 1 and 0.001 µg/g for FB 2 and FB 3 .
The BigDye Terminator kit v. 3.1 (Life Technologies, Carlsbad, CA, USA) was used for fluorescent labeling according to the manufacturer's instructions. DNA fragments were purified using alkaline phosphatase and exonuclease I (Thermo Fisher Scientific)) and precipitated using ice-cold 96% ethanol (Sigma Aldrich, St. Louis, MO, USA). Sequence reading was performed using Applied Biosystems equipment. Sequence reads were analyzed using BioEdit software [63] and aligned using MEGA5 software package [64] using Maximum Parsimony heuristics with standard settings. Based on FUM1 sequences, the most parsimonious tree was calculated (bootstrap test with 1000 replications).

Statistical Analysis
To analyze the in vitro fumonisin biosynthesis within each country of origin, total fumonisin content was submitted to ANOVA by allowing a different standard deviation per strain to comply with heteroscedasticity. Generalized least-squares were used for model fitting, as implemented in the gls() function of the nlme package [65] within the R statistical environment [66]. Heteroscedastic Welch's t-tests were used for pairwise comparisons of strains, within country [67].

Identity Confirmation of F. verticillioides
DNA extracted from the 90 F. verticillioides strains was subject to PCR assays using the species-specific primer pair VERT1/VERT2. As expected, a single fragment of 800 bp amplified in all the samples, thus confirming their identity as F. verticillioides.

Fumonisin Biosynthesis by F. verticillioides In Vitro
Data on the in vitro biosynthesis of FB 1 , FB 2 and FB 3 with the calculation of total fumonisins (sum of FB 1 , FB 2 and FB 3 ) by the 90 F. verticillioides strains are summarized in Table 1.
In general, this analysis revealed that 80% (n = 71) of the F. verticillioides strains investigated in this study were able to produce fumonisins at variable levels, while the remaining 20% (n = 19) showed undetectable levels (not detected; nd) of fumonisins and were considered, in this experimental condition, as non-producing strains.
In most cases, considering all producing strains (n = 71), differences in fumonisin production were detected among the strains isolated in the same country.
Taking into account all fumonisin-producing strains of each country analyzed in this study, differences in total fumonisin biosynthesis among countries were also detected ( Figure 3). In particular, the Spanish strains used in this study showed a significantly higher total fumonisin production (average 14.01 µg/g) than the Egyptian ones (average 3.98 µg/g) (p = 0.02). Also, the total fumonisin productions detected for the Italian (average 9.98 µg/g), Tunisian (average 5.36 µg/g) and Iranian (average 6.79 µg/g) strains were higher than the Egyptian ones and lower than the Spanish ones, even if no significant differences were recorded (p > 0.46 and p > 0.47, respectively) ( Figure 3). analogues (FB1, FB2 and FB3) were simultaneously produced by 64% of positive Iranian strains (n = 9), while 4 out of 14 strains (29%) produced only FB1, and 1 out of 14 strains (7%) biosynthesized FB1 and FB2. The Iranian strain 89 showed a significantly higher total fumonisin biosynthesis than the other strains from the same country (p < 0.01), with the exception of strains 5 and 7 (p > 0.05).
Taking into account all fumonisin-producing strains of each country analyzed in this study, differences in total fumonisin biosynthesis among countries were also detected ( Figure 3). In particular, the Spanish strains used in this study showed a significantly higher total fumonisin production (average 14.01 μg/g) than the Egyptian ones (average 3.98 μg/g) (p = 0.02). Also, the total fumonisin productions detected for the Italian (average 9.98 μg/g), Tunisian (average 5.36 μg/g) and Iranian (average 6.79 μg/g) strains were higher than the Egyptian ones and lower than the Spanish ones, even if no significant differences were recorded (p > 0.46 and p > 0.47, respectively) ( Figure 3).

Genetic Structure and Variability of F. verticillioides Populations
We sequenced a portion of a divergent FUM1 gene to evaluate the diversity among the five populations of F. verticillioides originating from various countries. All strains amplified DNA fragments of about 1100 bp in length. Additionally, the FUM6-FUM7 (ca. 550 bp) and FUM7-FUM8 (ca. 500 bp) intergenic regions were sequenced using the primers described previously [47].
The sequences were aligned, the ends trimmed manually using MEGA 5 software, and dendrograms of similarities were calculated. Interestingly, the intergenic regions did not show polymorphisms, which was rather unexpected, since these regions normally accumulated more point mutations than the coding regions. However, this means that the F. verticillioides strains characterized in this study, even if originating from different countries, were basically uniform (results not shown).
Therefore, only slightly more polymorphic FUM1 sequences were analyzed and shown ( Figure  4). Apparently, neither geographical origin nor fumonisin production ability were correlated to the genetic diversity of the strain set, as almost all of them grouped together. Only four strains from Egypt (F10, F12, F13 and F36) were distinguished from the remaining strains at a bootstrap value of 60, including our five reference sequences [61] and NCBI GenBank-deposited FUM cluster sequences (AF155773) reported by Proctor et al. [45]. Average of total fumonisins (µg/g) biosynthesized by Fusarium verticillioides fumonisin-producing strains isolated from maize kernels harvested in each of the five countries analyzed in this study. Means with different letters are significantly different (p < 0.05).

Genetic Structure and Variability of F. verticillioides Populations
We sequenced a portion of a divergent FUM1 gene to evaluate the diversity among the five populations of F. verticillioides originating from various countries. All strains amplified DNA fragments of about 1100 bp in length. Additionally, the FUM6-FUM7 (ca. 550 bp) and FUM7-FUM8 (ca. 500 bp) intergenic regions were sequenced using the primers described previously [47].
The sequences were aligned, the ends trimmed manually using MEGA 5 software, and dendrograms of similarities were calculated. Interestingly, the intergenic regions did not show polymorphisms, which was rather unexpected, since these regions normally accumulated more point mutations than the coding regions. However, this means that the F. verticillioides strains characterized in this study, even if originating from different countries, were basically uniform (results not shown).
Therefore, only slightly more polymorphic FUM1 sequences were analyzed and shown ( Figure 4). Apparently, neither geographical origin nor fumonisin production ability were correlated to the genetic diversity of the strain set, as almost all of them grouped together. Only four strains from Egypt (F10, F12, F13 and F36) were distinguished from the remaining strains at a bootstrap value of 60, including our five reference sequences [61] and NCBI GenBank-deposited FUM cluster sequences (AF155773) reported by Proctor et al. [45].

Discussion
This study was aimed at investigating the different ability of selected F. verticillioides strains isolated from maize kernels harvested in five Mediterranean countries to in vitro biosynthesize fumonisins as well as at characterizing their genetic structure to assess possible variabilities among them. So far, various studies have been conducted to analyze the ability of different F. verticillioides strains from diverse geographic areas to biosynthesize fumonisins. In several investigations, a large percentage of strains able to produce detectable amounts of these mycotoxins were usually found. However, the presence of strains that were not able to biosynthesize measurable levels of fumonisins was also reported. In this research, the majority of the strains isolated from maize grains in Italy, Spain, Tunisia and Iran, analyzed in this study, produced detectable levels of fumonisins (91%, 100%, 94% and 94% respectively; Figure 2), while the remaining part showed a lack of ability to produce measurable amounts of these mycotoxins. Similar percentages of fumonisin-producing strains (> 80%) were also detected in other F. verticillioides populations isolated from maize in Croatia [68], Spain [15,69], Italy [50], Iran [22], Egypt [17], Brazil [41,44,49], Korea [70], USA [71], Argentina [55,72] and from durum wheat in Argentina [2].
Conversely, in this study, only 46% of the analyzed Egyptian strains showed the ability to biosynthesize detectable amounts of fumonisins ( Figure 2). Similarly to other studies, low incidences of producing strains were also recorded in other F. verticillioides populations such as those isolated from maize in Croatia (55%) [73], Taiwan (66%) [74] and Spain (36%) [14].
In general, the producing strains analyzed in this study biosynthesized fumonisin analogues following the "typical" gradient: FB 1 > FB 2 > FB 3 . A predominance of FB 1 compared to the other analyzed fumonisin analogues was recovered also in other F. verticillioides populations such as those isolated from maize in Spain [15,75], Italy [76], Iran [22], Brazil [44,49], Argentina [55,72], Egypt [17], South Korea and South Africa [39]. In this study, no F. verticillioides strains producing more FB 2 or FB 3 than FB 1 were recorded. Conversely, these types of strains were observed in F. verticillioides populations isolated from durum wheat in Argentina [2] and from maize and sorghum cultivated in the United States [77].
As known, fumonisin production within the F. verticillioides species could quantitatively vary due to the different biosynthetic ability of the different strains [24,40]. Also in this study, variability of fumonisin production among strains isolated in the same country was found, highlighting that mycotoxigenic diversity occurred within the five investigated F. verticillioides populations. Variability among F. verticillioides strains isolated from maize in the same country was commonly detected in many surveys in other parts of the world [2,8,15,17,22,44,49,55,[73][74][75].
Variability in fumonisin production was also recorded among F. verticillioides strains isolated from different countries [30,39,71]. Also in this study, differences in fumonisin production among strains of different geographic origin were detected. In particular, the Spanish and Egyptian strains analyzed in this study showed a high level of mycotoxigenic variability, being the populations with the highest and the lowest fumonisin productions, respectively.
Interestingly, these two populations were also those with the highest and lowest percentages of fumonisin-producing (Spain) and non-producing (Egypt) strains. Conversely, the other three investigated populations of F. verticillioides (isolated from Italy, Tunisia and Iran) considered in this study did not show a significant variability of fumonisin production. In agreement with the results of Vogelgsang et al. [78], it is important to consider that in vitro results cannot be fully extrapolated to in vivo conditions because there are several factors influencing Fusarium infections and secondary metabolite production in the field. However, in vitro results could provide important information, which may be useful to understand intra-population variability within a single country as well as inter-population variability among different countries.
In this study, the mycotoxigenic characterization of F. verticillioides strains from different geographic origins was coupled to the study of the genetic structure of these populations. The genetic diversity of F. verticillioides has been studied using multiple techniques, including AFLP and RAPD methods [50,53,79].
Recently, however, direct sequencing of specific genomic regions has become more popular because of its high discrimination power and accuracy. The FUM1 gene has already beeeen proven to be useful to assess species diversity inside the FFSC, serving as a source of phylogenetic and chemotypic markers [47], showing often higher levels of polymorphisms than constitutively expressed genes [e.g., beta tubulin (tub2) or translation elongation factor 1α (tef-1α)].
Our previous studies suggested there might be high levels of intraspecific genetic uniformity inside F. verticillioides populations, particularly when compared to the high diversity of the closely related species F. proliferatum [61,62,80,81]. The use of the FUM1 gene sequence analysis allowed for discrimination of subpopulations likely related to the host species of origin. We assumed that a similar rule would be valid for F. verticillioides; therefore, we added some pea-and pineapple-derived strains to the analysis (Figure 4). It was also possible that geographical differences between populations would become visible.
However, in the present study we could not confirm this hypothesis. In fact, this was in accordance to previous findings, which did not reveal significant differences between F. verticillioides strains from different hosts [61]. This was also confirmed by the sequence analysis of the intergenic regions between FUM6 and FUM7 as well as FUM7 and FUM8 genes (results not shown), which were previously used for polymorphism screening [47]. The most likely explanation for this situation may be the endophytic type of growth observed for this pathogen in maize, which combined with the extensive seed material transfer between countries and continents made the population uniform across the world. Another possibility is that FUM cluster integrity and structure undergoes much more strict selection pressure in F. verticillioides than in F. proliferatum. This may implicate that fumonisin production by F. verticillioides is more essential to complete its life cycle than it is for F. proliferatum. This issue was already reported by Glenn et al. [82] but never confirmed for F. proliferatum.
The only outlier obtained in this study was a group of four strains (F10, F12, F13 and F36) isolated from Egypt (Figure 4), which was distinct from the remaining strains. Only one of these strains (F13) produced fumonisins in detectable amounts (Table 1). They should be further studied to explain their genetic diversity.

Conclusions
In this study, we analyzed fumonisin production as well as genetic structures of five F. verticillioides populations isolated from maize kernels in five Mediterranean countries.
The characterization of a selected number of strains per country does not allow a general conclusion to be drawn at the country level; however, the results obtained in these experimental conditions highlighted: (i) the presence of an Egyptian population which differed from the others for its low percentage of fumonisin-producing strains; (ii) the presence of significant differences in fumonisin production within the strains isolated in each of the surveyed countries and, in some cases, also among populations isolated from different countries; (iii) the high level of genetic uniformity inside the populations analyzed; (iv) the general absence of correlation between geographical origin and/or fumonisin production ability with the genetic diversity of the strain set; (v) the presence of four Egyptian strains that were distinguished from the other strains at a bootstrap value of 60.