Fusarium Secondary Metabolite Content in Naturally Produced and Artiﬁcially Provoked FHB Pressure in Winter Wheat

: Fusarium head blight (FHB) is an important disease of wheat and production of mycotoxins makes it a major threat in most wheat-producing areas worldwide. This study aimed to identify the impact of epidemic FHB conditions (usage of artiﬁcial Fusarium inoculation) on mycotoxin levels in unprocessed wheat. Fusarium levels were monitored at two locations in two treatments (natural infection and inoculation with Fusarium graminearum and F. culmorum ) where 13 mycotoxins were evaluated by LC/MS-MS in six winter wheat varieties. Due to favorable conditions for infection with Fusarium fungi during the ﬂowering period at location Tovarnik, wheat varieties had higher disease severity and increased mycotoxin accumulation, compared to Osijek. The most abundant mycotoxins in treatment with inoculation with Fusarium fungi were deoxynivalenol (DON), culmorin (CUL) and hydroxyculmorins. In treatment with natural infection, DON did not exceed maximum limits set by EU. Varieties with lower initial resistance accumulated DON even in naturally infected samples at Tovarnik. These results highlighted the impact of environment variation in the production of Fusarium mycotoxins where FHB initial resistance had a higher impact on the accumulation of mycotoxins than general resistance. Furthermore, wheat samples with higher DON concentration also contained elevated levels of CUL and hydroxyculmorins, showing that CUL can have a possible role in Fusarium virulence. The FHB evaluations provide important information about the genetic resistance of wheat varieties, as well as risk assessment considering mycotoxin accumulation in epidemic conditions.


Introduction
Wheat is one of the major staple and one of the "big three" cereal crops with an annual worldwide production of over 600 million tons [1]. During the period of anthesis, the plant is the most susceptible to diseases that affect wheat heads and one of the main concerns is Fusarium head blight (FHB) caused by fungi of the genus Fusarium. The disease can result in direct and indirect economic losses thus causing reduced grain yield and quality, as well as production of mycotoxins [2].
A wide range of factors play different roles in the growth, survival and dissemination of the fungus and thus can influence disease severity and mycotoxin production [3]. Primarily, the aggressiveness of Fusarium species and accumulation of mycotoxins is determined by wheat genetic variation [4]. The presence and incidence of different Fusarium

Plant Material and Field Trials
The study was conducted in vegetative season 2019/2020 at two locations, Osijek (45 • 32 N, 18 • 44 E) and Tovarnik (45 • 10 N, 19 • 09 E), Croatia. The soil types in these two regions are different with eutric cambisol present at Osijek and black soil chernozem at Tovarnik. During the period of flowering, the average precipitation was 1.7 mm at Osijek (Supplementary Table S1, Figure S1) and 2.3 mm at Tovarnik (Supplementary Table S2, Figure S2), and the average temperature was 15.3 • C at Osijek (Supplementary Table S3, Figure S1) and 15.6 • C at Tovarnik (Supplementary Table S4, Figure S2). The experimental plot area was 7.56 m 2 , where treatments (naturally infected and artificially inoculated) were replicated in two plots. In each treatment, same winter wheat varieties (El Nino, Galloper, Tika Taka, Vulkan, Kraljica and Golubica) originated from Agricultural institute Osijek, were used. The seed was treated with Vitavax 200 FF (thiram + carboxin) at a rate of 200 g Vitavax for 100 kg of seeds in order to control seed-borne diseases. Fungicides were excluded in both treatments and in the both investigated environments. Weed control was conducted with a herbicide at wheat tillering (GS 31). Insecticides were sprayed in the spring of the growing season. Fertilization was in proportions N:P:K 130:100:120 kg ha −1 . The grains were taken by harvesting the whole plot with a Wintersteiger cereal plot combine-harvester.

Inoculum Preparation and Inoculation Procedure
The Fusarium species used in this experiment were the two most prevalent causal agents of Fusarium head blight: Fusarium graminearum (PIO 31), isolated from the winter wheat collected in East Croatia, and F. culmorum (IFA 104) obtained from IFA-Tulln, Austria. Conidial inoculum of Fusarium spp. were produced by a mixture of wheat and oat grains (3:1 by volume). Conidial concentrations of both fungi were determined using a hemocytometer (Bürker-Türk, Hecht Assistent) and were set to 10 × 10 4 mL −1 . The 100 mL of inoculum was sprayed with sprayers on an area of m 2 at the flowering stage. One treatment was grown according to standard agronomical practice with no usage of fungicide and without misting treatment, while another treatment was subjected to two inoculation events using a tractor-back (Osijek) and hand sprayer (Tovarnik) with Fusarium spp. at the time of flowering (Zadok's scale 65) [27] (Figure 1a). Misting was provided by spraying with a tractor back-sprayer on several occasions.

Fusarium General Resistance and Type I Resistance
The percentage of bleached spikelets (Figure 1b) per plot and initial infection w estimated on days 10, 14, 18, 22, and 26 after inoculation according to a linear scale 100%). Based on the percentages, the area under the disease progress curve (AUDPC) general resistance and type I FHB resistance of wheat varieties was calculated accordi to formula: where Yi is percentage of visibly infected spikelets (Yi/100) at the ith observation, Xi is d of the ith observation and n is total number of observations.
Confirmation of positive analyte identification was obtained by the acquisition of tw MRMs per analyte (with the exception of moniliformin and 3-nitropropionic acid that hibit only one fragment ion), which yielded 4.0 identification points according to comm sion decision [29]. In addition, the liquid chromatography retention time and the intens ratio of the two MRM transitions agreed with the related values of an authentic standa within 0.03 min and 30% rel., respectively. Quantification was performed via external c ibration using serial dilutions of a multi-analyte stock solution. Results were corrected apparent recoveries obtained for wheat [28] (Supplementary Tables S5-S9). The accura

Fusarium General Resistance and Type I Resistance
The percentage of bleached spikelets (Figure 1b) per plot and initial infection were estimated on days 10, 14, 18, 22, and 26 after inoculation according to a linear scale (0-100%). Based on the percentages, the area under the disease progress curve (AUDPC) for general resistance and type I FHB resistance of wheat varieties was calculated according to formula: where Yi is percentage of visibly infected spikelets (Yi/100) at the ith observation, Xi is day of the ith observation and n is total number of observations.
Confirmation of positive analyte identification was obtained by the acquisition of two MRMs per analyte (with the exception of moniliformin and 3-nitropropionic acid that exhibit only one fragment ion), which yielded 4.0 identification points according to commission decision [29]. In addition, the liquid chromatography retention time and the intensity ratio of the two MRM transitions agreed with the related values of an authentic standard within 0.03 min and 30% rel., respectively. Quantification was performed via external calibration using serial dilutions of a multi-analyte stock solution. Results were corrected for apparent recoveries obtained for wheat [28] (Supplementary Tables S5-S9). The accuracy of the method is verified on a continuous basis by regular participation in proficiency testing scheme organized by BIPEA (Gennevilliers, France).

Statistical Analysis
The data were subjected to analysis of variance (ANOVA) using an appropriate model by Statistica version 12.0 (Statsoft Inc., Tulsa, OK, USA). To estimate disease progress, the AUDPC was used to combine multiple observations from five data points (different dates) into a single value. For correlation analyses, Spearman's coefficient was applied, shown in the supplementary file.

Fusarium General Resistance and Type I Resistance
The FHB symptoms varied among locations, where at Tovarnik there was higher area under the disease progress curve (AUDPC) for general resistance on average for three varieties (El Nino, Galloper and Tika Taka), compared to Osijek. The highest Type I and general resistance at Osijek had Galloper, as well as general resistance at Tovarnik. The lower AUDPC for initial resistance (higher Type I resistance) had Vulkan and Kraljica at Tovarnik. The highest AUDPC for Type I resistance was recorded for El Nino (AUDPC 421) at Tovarnik, followed by Golubica at Osijek (AUDPC 222) ( Table 1).

Nivalenol and Zearalenone
Concentrations of nivalenol (NIV) and zearalenone (ZEN) were lower than the rest of the Fusarium metabolites studied at both locations. At Osijek, NIV was found only in artificially inoculated samples of most FHB susceptible varieties (El Nino, Tika Taka and Golubica) where the levels were 29 μg kg −1 , 31 μg kg −1 and 37 μg kg −1 , respectively ( Figure  3a). At Tovarnik, NIV was found in all artificially inoculated samples and the concentrations ranged from 24 μg kg −1 in Galloper to 105 μg kg −1 in Golubica with an overall mean

Nivalenol and Zearalenone
Concentrations of nivalenol (NIV) and zearalenone (ZEN) were lower than the rest of the Fusarium metabolites studied at both locations. At Osijek, NIV was found only in artificially inoculated samples of most FHB susceptible varieties (El Nino, Tika Taka and Golubica) where the levels were 29 µg kg −1 , 31 µg kg −1 and 37 µg kg −1 , respectively (Figure 3a). At Tovarnik, NIV was found in all artificially inoculated samples and the concentrations ranged from 24 µg kg −1 in Galloper to 105 µg kg −1 in Golubica with an overall mean of 44 µg kg −1 (Figure 3b). In naturally infected samples it was detected only in Golubica (70 µg kg −1 ).  (Figure 3b). In naturally infected samples it was detected only in Golubica (70 μg kg −1 ). All artificially inoculated samples at both locations were contaminated with ZEN. None of the naturally infected samples contained ZEN except the El Nino variety at Tovarnik and its concentration was 1 μg kg −1 . The levels of ZEN measured at artificially inoculated six varieties at Osijek were 1 μg kg −1 in Galloper, 9 μg kg −1 in Golubica, 9 μg kg −1 in El Nino, 12 μg kg −1 in Vulkan, 16

Culmorin, 15-Hydroxyculmorin, 15-Hydroxyculmoron and 5-Hydroxyculmorin
El Nino, Tika Taka and Golubica accumulated culmorin (CUL) and its derivatives in much higher concentrations than other varieties. At Osijek, concentrations of CUL were elevated in the artificially inoculated compared to naturally infected samples. The highest amount of CUL was recorded in an artificially inoculated Golubica variety (29,100 μg kg −1 ). Other concentrations ranged from 7810 μg kg −1 in Kraljica to 13,100 μg kg −1 in Vulkan. In naturally infected samples CUL was found only in Tika Taka variety at concentration of 220 μg kg −1 (Figure 4a). At Tovarnik, concentrations of CUL were elevated even in naturally infected samples. The highest concentration was found in El Nino variety (1000 El Nino, Tika Taka and Golubica accumulated culmorin (CUL) and its derivatives in much higher concentrations than other varieties. At Osijek, concentrations of CUL were elevated in the artificially inoculated compared to naturally infected samples. The highest amount of CUL was recorded in an artificially inoculated Golubica variety (29,100 µg kg −1 ). Other concentrations ranged from 7810 µg kg −1 in Kraljica to 13,100 µg kg −1 in Vulkan. In naturally infected samples CUL was found only in Tika Taka variety at concentration of 220 µg kg −1 (Figure 4a). At Tovarnik, concentrations of CUL were elevated even in naturally infected samples. The highest concentration was found in El Nino variety (1000 µg kg −1 ) and the lowest was recorded in Tika Taka variety (62 µg kg −1 ), while in Kraljica it was not found. In artificially inoculated samples concentrations ranged from 6100 µg kg −1 in Galloper to 14,300 µg kg −1 in Golubica with an overall mean of 11,400 µg kg −1 (Figure 4b).

ANOVA and Correlation Analysis
Analysis of variance (ANOVA) revealed significant differences in 13 investigated mycotoxins among two treatments (p < 0.001, p < 0.01). Non-significant difference was found between varieties and locations for the most mycotoxins, except for 3-acetyldeoxynivalenol (3ADON), nivalenol (NIV) and fusarin C between locations. Moreover, significant differences were recorded between varieties for NIV (p < 0.01) ( Table 2).

Discussion
The study has focused on the effect of epidemic FHB conditions on the production of a range of metabolites originated either from the naturally contaminating mycobiota, or from inoculation by Fusarium graminearum and F. culmorum isolates. Furthermore, this study extended the investigation beyond the well-known mycotoxins to a range of fungal secondary metabolites accumulated in wheat grains. According to previous findings, fungal secondary metabolites commonly found in wheat grains are deoxynivalenol (DON), a type B trichothecene, and zearalenone (ZEN), while predominant species producing these mycotoxins in Europe are found to be F. graminearum, F. culmorum and F. avenaceum [30]. Given the fact that F. graminearum widely occurs in Europe, it is also predominantly found in Croatia [31]. However, many Fusarium metabolites are far less investigated than DON and ZEN [32] and therefore are not subject to legislation and regular monitoring. Both emerging mycotoxins and modified forms represent a new issue for food contamination [33].
The mycotoxigenic fungi produce several secondary metabolites at the time [34]. Therefore, this study reports the occurrence of 13 Fusarium metabolites and their concentrations in the wheat grains of six artificially inoculated as well as in naturally infected (field grown) winter wheat varieties (El Nino, Galloper, Tika Taka, Vulkan, Kraljica and Golubica). The combined use of resistant wheat varieties, fungicides, and specific management practices can reduce part of the Fusarium head blight (FHB) losses [35]. Therefore, the impact of fungicides in the current research was omitted, as well as the influence of management practice, as field experiments were done according to standard agronomical procedures. Considering that precipitation levels between two locations at which the experiment was held differed with Tovarnik having a higher precipitation rate and higher temperatures, levels of Fusarium metabolites studied were higher at Tovarnik, compared to Osijek. This is in accordance with previous reports where warm and humid conditions at and shortly after anthesis favour FHB [36]. Humidity determines the intensity of the disease, while precipitation determines inoculum levels [37]. In addition, wheat varieties in treatment with inoculation with Fusarium fungi were evaluated for Type I (initial) and general FHB resistance in the field conditions prior to harvest by calculating AUDPC.

Deoxynivalenol, Deoxynivalenol-3-glucoside and 3-Acetyldeoxynivalenol
Taking into account only proved Fusarium mycotoxins and not all metabolites studied, the current research was in accordance with previous studies which reported that DON is the most abundant mycotoxin in wheat grains [38,39]. In this study, varieties with higher initial susceptibility (El Nino, Tika Taka and Golubica) accumulated DON even in naturally infected samples. The average level of DON in naturally infected winter wheat plants (control samples) at location Tovarnik did not exceed the legal limit set by EU (1 250 µg kg −1 ) for unprocessed cereals [40] while at Osijek in naturally infected samples it was not found. The same results were previously reported where in randomly selected wheat samples from natural infection in Croatia, DON levels were below this threshold [41]. However, in the current research FHB artificially inoculated plants exceeded maximum levels for DON contamination 10-fold at Osijek and 15-fold at Tovarnik. As DON significantly correlated with few investigated mycotoxins, it can be concluded that DON content can be used in the selection for FHB resistance and could potentially participate in lowering total toxicity.
Deoxynivalenol-3-glucoside (D3G) is one of the main DON metabolites known as "modified mycotoxin" [42]. After ingestion it can be hydrolysed to DON [43,44]. D3G was present in Fusarium infected samples at both locations, while in naturally infected plants it was observed in susceptible varieties El Nino, Tika Taka and Golubica only at Tovarnik. Concentrations observed in the current research were similar to previous studies reporting the occurrence of DON and D3G in durum wheat in Italy [45]. Previously it was concluded that D3G usually comes in lower concentrations, compared to concentrations of DON [46].
3-acetyldeoxynivalenol (3ADON) was observed in Fusarium infected samples at Osijek as well as at Tovarnik. At Tovarnik it was also found in naturally infected samples in susceptible El Nino variety. In some researches, 3ADON was among the most abundant mycotoxins [47,48], which was not the case in the current research.
Correlations between DON and 3ADON were highly significant at Tovarnik which is in accordance with the previous research [49] where DON highly correlated with 3ADON in barley samples. Also, this correlation was expected in the current research as DONproducing strains with the 3-acetylated precursor are common in Europe [50]. The more pronounced FHB symptoms the higher correlations between those mycotoxins occurred, where higher 3ADON production may be associated with elevated DON content. It was concluded that comparatively higher levels of gene expression may contribute to the higher levels of DON produced by 3ADON strains in infected grains [51]. Although there was not any significant correlation between DON and general and Type I resistance, it can be assumed that high level of DON occurred in varieties with different level of symptoms, as it was evidenced in previous studies [52] reporting that the occurrence of high humidity post-anthesis produced late infections, with a high level of DON, but low level of FHB symptoms.

Nivalenol and Zearalenone
The obtained results showed that nivalenol (NIV) at Osijek was represented only in varieties with high initial susceptibility (El Nino, Tika Taka and Golubica) in artificially inoculated samples. At Tovarnik, it was found in all Fusarium infected plants while in naturally infected samples it was observed only in susceptible variety Golubica. This is in accordance with previous studies [53] where NIV was found only in one naturally infected wheat sample. This was expected for Fusarium infected samples as both F. graminearum and F. culmorum produce NIV at all tested temperatures between 10 and 30 ºC [54]. In the previous research the highest concentration of NIV in the organic cereal samples was 106 µg kg −1 [55], while the concentration of NIV in the current study was lower in the Fusarium infected as well as in the naturally infected samples.
Zearalenone (ZEN) concentrations were the lowest of all Fusarium metabolites studied at both locations and did not exceed risk threshold levels of 100 µg kg −1 for unprocessed cereals other than maize [40] in artificially inoculated nor in naturally infected samples. Our findings are in accordance with previous studies which showed that concentrations of ZEN found in wheat mostly do not exceed 50 µg kg −1 [32,56]. As expected, at Osijek ZEN was found only in inoculated samples while at Tovarnik in naturally infected plants it was observed only in susceptible variety El Nino in negligible concentration. Although there were not any significant correlations between DON and ZEN, it was previously concluded that ZEN is often co-produced with DON by Fusarium spp. such as F. graminearum [57]. For example, in another study where large European screening for Fusarium mycotoxins was obtained, ZEN was found in 32% of approximately 5000 samples of cereal grains and products tested [58].

Culmorin, 15-Hydroxyculmorin, 15-Hydroxyculmoron and 5-Hydroxyculmorin
In the novel time, there are numerous evidence for culmorin (CUL) being an "emerging mycotoxin". It is confirmed that CUL can inhibit the reaction of glucuronidation and thus increase the toxicity of DON [15,22]. Wheat samples with higher DON concentration contained elevated levels of CUL which implies that CUL can have a possible role in Fusarium virulence. Previously, it was reported about interactions between CUL and DON and its toxicity in growing pigs [59]. Furthermore, synergism between CUL and trichothecenes in plants occurred [60]. In Fusarium infected samples varieties with higher initial susceptibility (El Nino, Tika Taka and Golubica) accumulated CUL in much higher concentrations than other varieties. In Norway there was a concentration of 100 µg kg −1 detected in wheat [61].
Except for CUL, in the analysed samples also occurred 15-hydroxyculmorin, 15hydroxyculmoron and 5-hydroxyculmorin. Our study was partially in accordance with other researches which showed that in samples artificially inoculated with F. culmorum 15-hydroxyculmorin was the most abundant derivative of CUL and 5-hydroxyculmorin second abundant [23] as in the current research at Osijek, while at Tovarnik this was opposite with the most abundant 5-hydroxyculmorin and second 15-hydroxyculmorin. Moreover, in the above-mentioned study in naturally infected samples, CUL was the major metabolite, while in the current research in naturally infected samples CUL and its derivatives were found only at Tovarnik where the major metabolite was 15-hydroxyculmorin. The current study is in accordance with research which reporting that in naturally infected samples of wheat in Croatia concentration of CUL and 15-hydroxyculmorin is at a similar range as the concentration of DON thus implying that DON is correlated with CUL and hydroxyculmorins [62]. In addition to that at Tovarnik, CUL was in a positive significant correlation with AUDPC for general resistance thus implying that CUL role in Fusarium virulence is more pronounced in increased FHB pressure.

Aurofusarin, Butenolide, Chrysogin and Fusarin C
Fusarium metabolites aurofusarin, butenolide, chrysogin and fusarin C observed in this study are recently discovered and therefore far less investigated than others [39]. Under increased FHB pressure, aurofusarin was observed in all varieties at both, Osijek and Tovarnik, while in naturally infected samples it was observed only in Tovarnik in susceptible variety El Nino. In naturally infected samples of wheat aurofusarin was found in wheat in levels up to 4200 µg kg −1 [63]. At Osijek, the average level of aurofusarin was in accordance with previous study [64], while at Tovarnik it was observed in much higher concentrations implying that there is a certain connection between the production of major Fusarium mycotoxins and pigments. There was an even higher concentration of aurofusarin detected up to 140,000 µg kg −1 in Italian samples of durum wheat [65] than in the current research at Tovarnik. Although previous studies report aurofusarin and rubrofusarin accompanied by one another [66], in the current research rubrofusarin was not observed. According to previous studies there is the genetic and biosynthetic origin of aurofusarin and both DON and ZEN [67].
Determined levels of butenolide and chrysogin were lower than those previously reported [64]. However, results are in accordance with the research where fusarin C and chrysogine concentrations were higher in the wheat with F. graminearum treatments in contrast to the naturally stored wheat [68]. Higher concentration of fusarin C (average level 40,042 mg kg −1 ), chrysogin (average level 39 mg kg −1 ) and butenolide (7300 mg kg −1 ) were obtained in durum wheat in the fields with natural infection [65], while aurofusarin (average level 76,875 mg kg −1 ) was at a similar level as in the present study. Butenolide and fusarin C were expected to be detected in the current research because both F. graminearum and F. culmorum have the ability to produce them [17]. Furthermore, current research revealed high positive correlation between chrysogin and aurofusarin, which was expected, as chrysogin is also pigment produced by Fusarium sp. [69].

Conclusions
This study performed at two different locations indicated that winter wheat samples with higher deoxynivalenol (DON) concentration contained elevated levels of culmorin (CUL) and hydroxyculmorins, showing that CUL can have a possible role in Fusarium virulence, which became more pronounced in elevated infection with Fusarium fungi. Since DON significantly correlated with few investigated metabolites, it can be assumed that DON can participate in lowering total toxicity. Furthermore, according to elevated aurofusarin levels, it is also assumed that there is a certain connection between the production of major Fusarium mycotoxins and pigments under increased FHB pressure. As the impact of some fungal secondary metabolites on food and feed safety, i.e., human and animal health is still unclear, it is of great importance to investigate their toxicity, as well as consequently regulate maximal allowed concentration for specific food and feed. Furthermore, possible synergistic effects between certain metabolites need to be investigated more closely as they could interact together thus giving total toxicity. An inevitable practical conclusion of this manuscript is also information about the genetic resistance of winter wheat varieties investigated which will be useful for future risk assessment, considering FHB pressure and consequently mycotoxin production.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/agronomy11112239/s1, Figure S1: Climate diagram for May 2020 at Osijek. Figure S2: Climate diagram for May 2020 at Tovarnik. Table S1: Daily precipitation (mm) during May, June and July at Osijek in 2020. Table S2: Daily precipitation (mm) during May, June and July at Tovarnik in 2020. Table S3: Daily temperatures ( • C) during May, June and July at Osijek in 2020. Table S4: Daily temperatures ( • C) during May, June and July at Tovarnik in 2020. Table S5: Apparent recovery values, LOD values and LOQ values for 13 Fusarium secondary metabolites analyzed. Table S6: Amounts of Fusarium secondary metabolites in naturally infected samples at Osijek. Table S7: Amounts of Fusarium secondary metabolites in samples inoculated with Fusarium fungi at Osijek. Table S8: Amounts of Fusarium secondary metabolites in naturally infected samples at Tovarnik. Table S9: Amounts of Fusarium secondary metabolites in samples inoculated with Fusarium fungi at Tovarnik. Table S10: Correlation analysis between metabolite accumulation and the area under the disease progress curve (AUDPC) for general and Type I (initial) resistance at Osijek. Table S11: Correlation analysis between metabolite accumulation and the area under the disease progress curve (AUDPC) for general and Type I (initial) resistance at Tovarnik.

Conflicts of Interest:
The authors declare no conflict of interest.