A Novel Population of Fusarium fujikuroi Isolated from Southeastern U.S. Winegrapes Reveals the Need to Re-Evaluate the Species’ Fumonisin Production

Mycotoxins pose a challenge to a safe food supply worldwide, and their threat is expected to worsen with our changing climate. The need for diligence is exemplified by the discovery of fumonisin B2 in wine, which joins ochratoxin A as a mycotoxin of concern in the grape-wine chain. To elucidate the mycotoxin risk in southeastern American wine, grape samples were collected from vineyards during harvest in 2013 and potentially mycotoxigenic fungi (Fusarium and Aspergillus) were isolated from the samples. Numerous Fusarium isolates were recovered and identified to the species level by comparison of translation elongation factor 1-α gene sequences to verified strains. Fusarium fujikuroi was the most abundant species recovered (239 isolates), followed by F. proliferatum (52), F. incarnatum-equiseti (14), F. oxysporum (7), F. concentricum (1), and F. solani (1). In vitro assays quantified fumonisin production for representative isolates via liquid chromatography-tandem mass spectrometry. Surprisingly, nearly all F. fujikuroi isolates produced fumonisins B1, B2, and B3 at levels comparable to both the F. proliferatum isolates and the positive control, Fusarium verticillioides. Such capacity for fumonisin production refutes the generally accepted notion that F. fujikuroi produces undetectable or low levels of fumonisins and provides evidence to reconsider this species as a mycotoxigenic threat to economically significant crops.


Introduction
The possibility exists that mycotoxins, toxins produced by fungi having adverse effects on humans and animals, can survive the winemaking process to contaminate bottled wine even after the fungi themselves are eliminated from the must. The mycotoxin ochratoxin A (OTA) was discovered in 1965 by van der Merwe et al. [1], and the mycotoxins fumonisin B1 (FB1) and fumonisin B2 (FB2) were identified in 1988 [2]. In 1996, Zimmerli and Dick discovered OTA in wine [3] and more recently, in 2010 FB2 was revealed as a wine contaminant [4,5]. OTA, a mycotoxin produced by black-spored Aspergillus species such as Aspergillus carbonarius and Aspergillus niger [6][7][8], and present in cereals, wine, and coffee, has either been shown to be or is potentially nephrotoxic, immunosuppressive, teratogenic, genotoxic, embryotoxic, and cytotoxic [9]. OTA is also a possible human carcinogen in Group 2B, with fellow members dichlorodiphenyltrichloroethane (DDT), lead, and chloroform [10]. The source of FB2 in wine has been attributed to A. niger [11,12]. This is in contrast to FB1 in maize, which is the most common fumonisin in that commodity and is produced mainly by Fusarium verticillioides [13]. FB1 is

Fungal Isolates Collected from Winegrapes
A total of three black-spored Aspergillus isolates were recovered from the winegrape samples, each one from a different vineyard. None of these isolates had the morphological characteristics of the main OTA producer, A. carbonarius, and thus were not further investigated for mycotoxin production. In contrast, a total of 314 Fusarium isolates were recovered from the winegrape samples using dichloran rose bengal chloramphenicol (DRBC) and 2-benzoxazolinone (BOA) selective media. Based on sequencing of the translation elongation factor 1-α (TEF) region and BLASTN searches of three nucleotide databases for species identification, F. fujikuroi was the most abundant species recovered (n = 239, 76.1% of isolates), followed by F. proliferatum (n = 52, 16.6%), F. incarnatum-equiseti (n = 14, 4.5%), F. oxysporum (n = 7, 2.2%), F. concentricum (n = 1, <1%), and F. solani (n = 1; <1%). F. fujikuroi isolates were recovered from every vineyard (n = 8) and every grape variety (n = 10) sampled (Table 1), whereas F. proliferatum was recovered from fewer vineyards (n = 3) and grape varieties (n = 4), and only from the states of North Carolina and Alabama. The importance of collecting more than one isolate from a single sample replicate was demonstrated, as multiple species were found in distinct replicates. To further support the determination that isolates were either F. fujikuroi or F. proliferatum, percent identity matrices were generated from ClustalW nucleotide alignments of winegrape isolates compared to multiple reference strains of each species. Identity of ≥97% was used to make species determinations, and F. fujikuroi was clearly distinguishable from F. proliferatum (Supplementary File). Also, due to the large number of isolates obtained from the winegrapes, only representative isolates from across vineyards, grape varieties, and sample replicates were deposited in the Agricultural Research Service (ARS) Culture Collection (NRRL), and the TEF sequences from these representatives were deposited in the National Center for Biotechnology Information (NCBI) GenBank (Table 2). Percent identity matrices of TEF alignments for those isolates not curated in NRRL and GenBank are presented in the Supplementary File as support of their species determination.

In Vitro Fumonisin Production Analysis
A representative group of F. proliferatum and F. fujikuroi winegrape isolates were further explored for their fumonisin producing capabilities in vitro using a rice-based assay ( Table 2). Nearly all tested F. proliferatum isolates (20/22) produced fumonisins in the range of 1125-4653 ppm (µg/g) total fumonisins (FB1, FB2, and FB3). Average total fumonisins for F. proliferatum isolates was 1922 ppm. All F. fujikuroi isolates (50/50) produced fumonisins with an average total fumonisins of 1997 ppm (range of 1.09-3921 ppm). The average total fumonisins produced by the positive control, F. verticillioides, was 1675 ppm. No F. oxysporum isolates tested produced fumonisins in this study. There were no correlations with fumonisin production and state, vineyard, grape variety, or media isolation type (DRBC or BOA). Furthermore, isolates with the TEF SNP corresponding with non-production according to Suga et al. [26] did produce fumonisins in this study ( Table 2).   . The translation elongation factor (TEF) sequences of these curated isolates were deposited in the National Center for Biotechnology Information (NCBI) GenBank. Other winegrape isolates, indicated by a dash, were sequenced to confirm species identity, but sequences were not deposited in GenBank since the isolates were not curated in NRRL. n/a = not applicable to F. verticillioides positive controls; c Y = SNP (single nucleotide polymorphism in TEF region) identified by Suga et al. [26] to be a fumonisin producer. N = SNP identified to be a non-producer. n/a = the SNP applies to F. fujikuroi and is not applicable to other species; d Represents the average of a minimum of two independent experiments, each with three technical replicates. Unit = ppm = µg toxin per gram of dried rice. ND = none detected, limit of detection (LOD) = 0.013 ppb, and limit of quantification (LOQ) = 0.045 ppb; e ND = none detected, LOW = <500 ppm, MEDIUM = 500-1500 ppm, HIGH = >1500 ppm; f SLB 6100 is the only isolate tested which had discrepancy in production of FB2 between assays. In three of the four assay replicates (each tested in triplicate), this isolate produced no detectable FB2, but in one assay replicate it produced 1200 ppm average FB2.

Discussion
Whereas F. proliferatum is a well-known fumonisin producer, the general consensus in the Fusarium research community has been that F. fujikuroi produces little or no fumonisins. Surprisingly, all but two F. fujikuroi winegrape isolates tested with the rice assay produced significant levels of total fumonisins (greater than 400 ppm; either B1, B2, B3, or a combination). The levels of fumonisins produced by the F. fujikuroi winegrape isolates are comparable to the levels produced by the F. proliferatum winegrape isolates and to the F. verticillioides positive controls. This is the first report of a large population of F. fujikuroi isolates with the ability to produce high amounts of fumonisins.
Due to the rarity of a high fumonisin-producing F. fujikuroi population, further testing with verified F. fujikuroi and F. proliferatum isolates from the NRRL Culture Collection was conducted to determine if the rice assay method of in vitro fumonisin production, less common than the alternative cracked corn assay, was perhaps the cause of such production. One of these verified F. fujikuroi isolates, NRRL 31883, was isolated from a grape host in NY, USA. This one isolate, but none of the other NRRL F. fujikuroi reference strains we tested, produced high amounts of fumonisins similar to the southeastern winegrape isolates (Table 3). All other F. fujikuroi NRRL reference strains produced undetectable to low amounts of fumonisins (<5 ppm). The species identity of all NRRL isolates tested was verified through sequencing of the TEF region as described earlier. Therefore, the rice assay testing method was not the cause of the high fumonisin production by F. fujikuroi winegrape isolates, which represent a truly unique population for this species. Differential regulation of the fumonisin biosynthetic gene cluster within this population may be the basis for the enhanced production of the mycotoxin. The grape host or the associated vineyard pest management practices may create an environment conducive to activation of the biosynthetic genes, perhaps related to nitrogen availability, which was shown to affect expression of the gene cluster [31]. Further studies would be needed to address these possibilities. We do wish to note that our laboratory has conducted fumonisin testing of southeastern American wines, and to date no wine samples have contained fumonisins at a level which would pose a consumer health risk (unpublished data).
Comparing published results of fumonisin production by Fusarium isolates is difficult due to the great diversity amongst in vitro fumonisin testing assays. Methods include a variety of growth media (cracked corn, rice, liquid broth, agar plates), incubation periods, toxin extraction methods, and means of reporting the toxin levels (ppm as µg toxin per gram of corn, µg toxin per gram dry fungal mass, etc.). Therefore, a more accurate and comprehensive approach is to obtain standard strains from a well-curated repository, verify their identity through sequencing, and test these strains in-house alongside collected experimental isolates. To this end, 35 of the winegrape Fusarium isolates were deposited into the permanent ARS Culture Collection. The NRRL numbers of these winegrape isolates are listed in Table 2. These strains may be useful to other researchers for in-house comparison of fumonisin production and genetic studies related to better understanding the F. fujikuroi secondary metabolite profile and its regulation. Given the wide host range of Fusarium species, this newly discovered high fumonisin-producing population of F. fujikuroi in vineyards in the Southeast raises concern for fumonisin contamination in other crops as well. Closer attention should be paid to the mycotoxigenic potential of F. fujikuroi worldwide.

Collection of Winegrape Samples
Samples of red, V. vinifera winegrapes were obtained from eight commercial vineyards in the southeastern U.S. (AL, GA, NC) during harvest in 2013. This harvest represented a vintage with both late season rains and high disease pressure. Grape varieties sampled included Cabernet Sauvignon, Cabernet franc, Malbec, Merlot, Petit Verdot, Pinot noir, Sangiovese, Tannat, and Touriga Nacional. One additional vineyard, planted with the non-vinifera variety of Norton, was also sampled. Up to three different grape varieties were sampled per vineyard. The majority of grape varieties sampled in each vineyard consisted of six clusters in total, each cluster randomly selected from a different non-adjacent grapevine on an individual vineyard row, forming up to three sample replicates of two clusters each. In total, 56 sample replicates were collected and processed. All clusters were brought to the laboratory on ice in re-sealable plastic bags and prepared for further analysis within 12 h from being cut from the vine. Stems were removed from clusters and grapes were homogenized (Polytron PT 10-35, Brinkmann Instruments, Thermo Fisher Scientific, Waltham, MA, USA) into a must slurry. The grapes were not surface sterilized in order to recover greater mycoflora diversity and to allow recovery of ochratoxigenic A. carbonarius or other black-spored Aspergillus species that typically sporulate on the grape berry exterior [15]. A sterile glycerin (Columbus Chemical Industries, Inc., Columbus, OH, USA) solution was added to grape must to a final concentration of 4%, and the must solution was frozen at −80 • C until further analysis.

Fungal Isolation
Thawed grape must sample replicates (1-3 per variety per vineyard) were serially diluted in sterile reverse osmosis (RO) water with 0.10% Tween ® 20 (Fisher Scientific, Fair Lawn, NJ, USA). Undiluted and serially diluted must samples were plated in duplicate onto two types of selective media: DRBC agar [32] and BOA agar [33]. Negative controls for each media type were included on every day of plating. Following a 5-7 day incubation period of DRBC plates at 30 • C and a 7-10 day incubation period of BOA plates at 27 • C, all potential black-spored Aspergillus and cottony-white Fusarium colonies were transferred onto potato dextrose agar (PDA) and incubated at 27 • C or 30 • C for purification and further identification. As Fusarium spp. were abundant, a maximum of 10 Fusarium colonies were isolated from each sample replicate and media type combination. Pure cultures of either Aspergillus or Fusarium were single-spore isolated via the dilution plating method on PDA [34]. Isolates were stored in 15% glycerin/0.1% Tween ® 20 at −80 • C for further analysis.

DNA Extraction and Isolate Identification
Single-spored Fusarium isolates were grown on PDA for 2-4 days at 27 • C and genomic DNA was extracted using the previously described thermolysis procedure [35]. The primers EF-1 (5 -GTTAAGAGGCGCGGTGTCGGTGTG-3 ) and EF-2 (5 -GGAAGTACCAGTGATCATGTT-3 ) [36], along with 1 µL of fungal genomic DNA and Taq Hot Start Quick-Load ® 2X Master Mix with Standard Buffer (New England BioLabs, Inc., Ipswich, MA, USA) were used to PCR-amplify a portion of the TEF gene for species identity [37]. PCR conditions were as follows: 94 • C for 30 s, followed by 39 cycles of 94 • C for 10 s, 55 • C for 10 s, and 68 • C for 45 s, with a final extension of 68 • C for 5 min. Gene amplification was verified by gel electrophoresis, then PCR products were sent to GenScript USA, Inc. (Piscataway, NJ, USA) for purification and DNA sequencing. Sequencing primers used were the same as PCR primers. Sequences were edited with Sequencher ® version 5.4.1 (Gene Codes Corporation, Ann Arbor, MI, USA). The TEF sequences were compared with verified strains in the Fusarium MLST (CBS-KNAW Fungal Biodiversity Centre, Utrecht, The Netherlands) [38], Fusarium-ID [39], and NCBI-BLAST [40] databases for species identification. Sequences from winegrape isolates were aligned to F. fujikuroi and F. proliferatum reference sequences from the databases using ClustalW implemented in Geneious analysis software (version 8.1.8, 2015, Biomatters Limited, Auckland, New Zealand) [41]. Percent nucleotide identity matrices were generated from the alignments, and an identity of ≥97% to reference isolates was used to designate winegrape isolates as either F. fujikuroi or F. proliferatum. Detailed phylogenetic characterization of these isolates will be published separately.

In Vitro Fumonisin Production Analysis
Following species identification, a subset (n = 73) of the total recovered potentially mycotoxigenic Fusarium winegrape isolates (n = 291) was selected for in vitro fumonisin analysis. This representative subset included both F. proliferatum (n = 22) and F. fujikuroi (n = 51) isolates from different states, vineyards, grape varieties, sample replicates, and media isolation types. Additionally, to test the Japanese isolate hypothesis that a SNP indicates fumonisin production or lack thereof, representative isolates with both producing and non-producing SNP types were included [26]. Four F. oxysporum winegrape isolates were assayed as well, though fumonisin B production is extremely rare for this species. Isolates of F. incarnatum-equiseti and F. concentricum, also from winegrapes, were included as negative controls. Additionally, strains of F. verticillioides, F. proliferatum, and F. fujikuroi were obtained from the USDA-ARS Culture Collection (NRRL) for inclusion in the fumonisin analysis (Peoria, IL, USA) [42]. Fumonisin-producing F. verticillioides strains (NRRL 20956, NRRL 20960, NRRL 20984) were used as positive controls, and one winegrape F. fujikuroi isolate (SLB 6326) that produced fumonisins in the first assay was used as an internal assay control for all subsequent assays. Therefore, each experimental run included testing potential fumonisin-producing Fusarium isolates from winegrapes and/or verified strains from the NRRL Culture Collection, a winegrape isolate negative control, a sterile water negative control, a F. verticillioides positive control, and an internal assay control. All isolates and strains analyzed, including controls, were tested in triplicate in each experiment. Assays were repeated on different days, and every winegrape isolate and verified strain was tested for fumonisin production in at least two independent experiments.
For fumonisin analysis, fungal isolates were grown on sterile rice using an assay modified from [27]. For each assay run, approximately 10 µL of a high-inoculum spore suspension from frozen single-spored stock cultures (−80 • C, in 15% glycerin/0.1% Tween ® 20) was used as inoculum for 50 mL of sterile potato dextrose broth (PDB) in a 125 mL Erlenmeyer shaking flask. Culture flasks were incubated for 3 days (±1 h) at 27 • C with shaking at 200 rpm in darkness. To a 20 mL glass scintillation vial, 3 g (±0.004 g) of rice (enriched long grain rice, Kroger, Athens, GA, USA) and 3 mL of reverse osmosis water were added. Vials were topped with foam plugs, autoclaved (121 • C, 30 min), and cooled to room temperature just prior to use. In triplicate, 250 µL of high inoculum spore suspension (~10 7 spores per mL) from incubated PDB flasks (or sterile water control) were added to each rice vial replicate. Inoculated rice vials were incubated for 7 days (±1 h) at 27 • C in darkness.
Following the incubation period on rice, 10 mL of a 1:1 acetonitrile (HPLC grade, Fisher, Fair Lawn, NJ, USA): distilled water extraction solution containing 5% formic acid (~98%, Sigma-Aldrich, St. Louis, MO, USA) was added to each vial. Samples were vigorously shaken and frozen to −20 • C for at least 10 h. Prior to LC-MS/MS analysis for fumonisins, rice cultures were broken apart and vials were rocked for 3 h (The Belly Dancer ® , speed 6.5, Stovall Life Science, Inc., Greensboro, NC, USA) to extract any fumonisins. The resulting extraction solution was centrifuge-filtered (costar ® Spin-X HPLC Micro Centrifuge Filter, 0.45 µm, nylon, Corning Inc., Corning, NY, USA), diluted (up to 10,000 fold), and analyzed via LC-MS/MS (Thermo LTQ XL). All analyses were multiple reaction monitoring (MRM) performed in positive ion mode, and the instrument was tuned using FB1 (PROMEC, Tygerberg, South Africa). An Imtakt Cadenza CW-C18 column (150 × 2 mm, 3 µm; Kyoto, Japan), maintained at 30 • C, was used for separations. Solvent A was distilled water with 3% acetonitrile plus 0.1% formic acid, and solvent B was acetonitrile with 3% distilled water plus 0.1% formic acid. The solvent program began at 30:70 A:B, and increased linearly to 100% B over 9 min, held at 100% B for 2 min, and then returned to starting conditions. Samples were run along with a blank control and standard solutions of fumonisins B1, B2, and B3 (1 ppb, 10 ppb, and 100 ppb in 30:70 acetonitrile:distilled water plus 0.1% formic acid; PROMEC, Tygerberg, South Africa). Analytes were identified by retention time, as compared to standards, and fragmentation patterns (Figure 1). Levels of fumonisins were quantified by external calibration of peak areas (the MS/MS total ion current) with standards, and a new calibration curve (limit of quantification (LOQ) = 0.045 ppb and limit of detection (LOD) = 0.013 ppb) was run with each sample set. Fumonisin amounts were reported as µg toxin per gram of dried rice (ppm).  Figure 1). Levels of fumonisins were quantified by external calibration of peak areas (the MS/MS total ion current) with standards, and a new calibration curve (limit of quantification (LOQ) = 0.045 ppb and limit of detection (LOD) = 0.013 ppb) was run with each sample set. Fumonisin amounts were reported as μg toxin per gram of dried rice (ppm).