Comparative Toxigenicity and Associated Mutagenicity of Aspergillus fumigatus and Aspergillus flavus Group Isolates Collected from the Agricultural Environment

The mutagenic patterns of A. flavus, A. parasiticus and A. fumigatus extracts were evaluated. These strains of toxigenic Aspergillus were collected from the agricultural environment. The Ames test was performed on Salmonella typhimurium strains TA98, TA100 and TA102, without and with S9mix (exogenous metabolic activation system). These data were compared with the mutagenicity of the corresponding pure mycotoxins tested alone or in reconstituted mixtures with equivalent concentrations, in order to investigate the potential interactions between these molecules and/or other natural metabolites. At least 3 mechanisms are involved in the mutagenic response of these aflatoxins: firstly, the formation of AFB1-8,9-epoxide upon addition of S9mix, secondly the likely formation of oxidative damage as indicated by significant responses in TA102, and thirdly, a direct mutagenicity observed for higher doses of some extracts or associated mycotoxins, which does not therefore involve exogenously activated intermediates. Besides the identified mycotoxins (AFB1, AFB2 and AFM1), additional “natural” compounds contribute to the global mutagenicity of the extracts. On the other hand, AFB2 and AFM1 modulate negatively the mutagenicity of AFB1 when mixed in binary or tertiary mixtures. Thus, the evaluation of the mutagenicity of “natural” mixtures is an integrated parameter that better reflects the potential impact of exposure to toxigenic Aspergilli.


Introduction
We have previously demonstrated in an agricultural environment (Normandy, France), that oilseed cakes and maize silage used for cattle food could be contaminated with Aspergilli, especially A. flavus, A. parasiticus and A. fumigatus, even in temperate/mild climate [1,2]. Moreover, bioaerosols resulting from breeding activities were also contaminated by airborne molds [3,4]. Among them, some species belonging to Aspergillus genus are well-known as potential producers of various mycotoxins such as aflatoxins, gliotoxin, verruculogen or fumagillin. The toxic effects of these purified mycotoxins

Mycotoxin Production
Toxigenic A. flavus and A. parasiticus strains both synthesize AFB 1 and, to a lesser extent, AFB 2 , whereas AFM 1 is naturally produced only by the toxigenic A. flavus (Table 1). Mycotoxin production is 20-to 40-fold higher for A. flavus as compared to A. parasiticus. None of these mycotoxins were detected for the so-called "non toxigenic" corresponding strains. The two strains of A. fumigatus were toxigenic, producing roughly comparable fumagillin and verruculogen levels, but they differ with regard to their gliotoxin synthesis potential (Table 1).  1 : aflatoxin M 1, GLIO: gliotoxin, VER: verruculogen, FUM: fumagillin; +: mycotoxigenic strain; −: non mycotoxigenic strain; Glio+/− : strain producing/not producing gliotoxin; <QL: below to the limit of quantification (2.5 ng/g for gliotoxin and 0.5 ng/g for other mycotoxins).

Mutagenicity of Fungal Extracts
Mutagenicity was evaluated using the Ames test in three Salmonella tester strains with and without exogenous metabolic activation system (S9mix). TA98 is sensitive to chemicals acting through frameshift mechanism, whereas TA100 and TA102 are sensitive to chemicals acting through base pair substitutions mechanism.

Aspergillus flavus Extracts
The mutagenic potential of the (Flav + ) extract is displayed in Table 2 (direct mutagenicity) and Table 3 (upon S9mix addition). Significant responses were observed in the three Salmonella tester strains, in absence as well as in presence of S9mix, except for strain TA102 for which no direct mutagenicity was observed.  Moreover, significant mutagenicities were systematically coupled with a clear dose-response relationship, and were dramatically increased in the presence of S9mix (Table 3). In contrast, no mutagenic response was observed from the (Flav−) extract.

Aspergillus parasiticus Extract
No mutagenic activity was detected for the (Para − ) extract (Tables 4 and 5). In contrast, upon S9mix addition, the (Para + ) extract was mutagenic in the three Salmonella tester strains (Table 5), with a dose-response relationship. Without S9mix (Table 6), a dose-dependent mutagenic response was obtained in TA98 only.

Aspergillus fumigatus Extract
There was no mutagenic activity observable in Salmonella tester strains TA98 and TA100 (Tables 6  and 7). In TA102, both extracts presented a low mutagenicity only upon addition of S9mix (Table 7).

Mutagenicity of the Mycotoxin Mixtures
Reconstituted mixtures were prepared from pure mycotoxins with concentrations corresponding to those measured in extracts. These binary and ternary mixtures were then tested for their mutagenic potential.

Mutagenicity of Pure Mycotoxins
Mutagenic activities of the 6 mycotoxins detected in the different extracts (Table 1) were also evaluated from pure standard solutions (Table 8). In a range of concentrations equivalent (or even wider) to those tested with the whole extracts, AFB 1 was the only one displaying a significant mutagenicity in TA98 and TA100, both without and with S9mix. In contrast, no mutagenic response was observed in TA102.

Discussion
Various Aspergilli species were isolated from our previous studies conducted in an agricultural environment. They differed considering both their ability to synthesize mycotoxins and the nature of these mycotoxins (Table 1). They originated from contaminated oilseed cakes or from bioaerosols sampled in dairy cattle sheds or during cattle feed distribution. Thus, spores and mycelium fragments of these Aspergilli species could be ingested by livestock and also inhaled by farmers during their daily work. In order to evaluate the potential hazard associated with occupational exposure to these fungal strains, we measured the mutagenic potential of total extracts obtained from MEA cultures. For this purpose, Salmonella typhimurium TA98, TA100 and TA102 tester strains were used (Ames test) without and with the exogenous metabolic fraction S9mix. The mutagenicity of the pure mycotoxins at the same concentrations as in the extracts was evaluated and compared in order to investigate the presence of additional mutagenic compounds in these extracts. It is noteworthy that, in the range of concentrations studied, corresponding to native production levels, AFB 1 was the sole mycotoxin displaying a significant mutagenicity when tested alone. Thus, extracts mutagenicity was discussed in comparison with that of an equivalent pure AFB 1 concentration (Table 9). AFB 1 and AFB 2 were found in the (Flav + ) extract and, to a lesser extent, in the (Para + ). It must be stressed that AFM 1 , which is mainly considered as a liver-hydroxylated metabolite of AFB 1 , was detected in the (Flav + ) extract. This direct production of AFM 1 by A. flavus was also recently reported by Uka et al. [16], but does not appear to have been commonly described.
A significant direct mutagenicity was observed in TA98 for (Flav + ) and (Para + ) extracts as well as for pure AFB 1 . Previously, Loarca-Piña et al. [17] reported a direct mutagenicity for AFB 1 , using the microsuspension assay. Compared with our study, they reported comparable mutagenic potencies for TA100 (0.5 vs. 0.54 revertants/ng) but somewhat higher for TA98 (0.4 vs 0.18 revertants/ng). In the present work, the responses observed for (Flav + ) and even more for (Para + ) extracts were largely enhanced (2-to 8-fold) compared with the same dose of pure AFB 1 (Table 9). Similarly, a higher direct mutagenicity was obtained in TA100 only for the (Flav + ) extract, compared to pure AFB 1 (Table 9).
Thus, these patterns indicate that AFB 1 is only one of the direct-and indirect-mutagenic compounds present in the total extracts. Moreover, the absence of mutagenicity for (Flav -) and (Para -) extracts could indicate that additional metabolites are closely related to the AFB 1 synthesis pathway. Recently, Uka et al. [16] highlighted the particularly high intra-species diversity of A. flavus and reported the large variety of secondary metabolites that can be produced. Using a metabolomic approach on 55 isolates of A. flavus, they identified more than 50 metabolites even if only half of the 55 strains were considered as aflatoxin producers. Beside aflatoxins (AFB 1 , AFB 2 and possibly AFB 2a , AFM 1 , AFG 1 , AFG 2 ) and their known precursors (sterigmatocystin and derived compounds), biocoumarins (aflavarin and derived compounds) with anti-insectan activity, isocoumarines (asperentin and derived compounds), anthraquinones (asparasone and derived compounds), non-ribosomal peptides (aspergillic acid and derived compounds) with antibiotic activity and indol-diterpenoids (aflavinines and aflatrem) with tremorgenic potential were identified. However, among these metabolites, only sterigmatocystin was reported as mutagenic in Ames test, both in TA100 (400 revertants/0.1 µg with S9mix and 150 revertants/0.1 µg without S9mix) and in TA98 (190 revertants/0.1 µg with S9mix and 40 revertants/0.1 µg without S9mix) [18].
Aflatoxicol displayed also a mutagenic potential in the presence of mammalian microsomes [19]. The production of aflatoxicol by A. flavus, as well as its possible interconversion with AFB 1 have been previously reported by Nakazato et al. [20,21].
Moreover, the mutagenic activities measured from (Flav + ) and (Para + ) extracts were always significantly higher than those obtained with the corresponding reconstituted mixtures of pure mycotoxins. These observations argue again in favour of additional components extracted from fungal cultures, and contributing to the overall mutagenic response.
The potential interactions between identified mycotoxins were also considered. Thus, the mutagenicity of reconstituted mixtures was compared to that of pure mycotoxins tested individually. Despite the fact that AFB 2 and AFM 1 were not mutagenic per se in our conditions, their association with AFB 1 was systematically linked to a significantly decreased mutagenicity of the corresponding binary or ternary mixtures, indicating a potential negative interaction between these structurally close molecules. A competitive mechanism during the S9 metabolic activation step could be evoked. Since AFB 2 and AFM 1 are structurally close to AFB 1 , they could also be substrate for cytochrome P450 enzymes (CYP), leading to a competition between these mycotoxins for the binding step to enzyme. Interaction could also occur later, once the 8,9-epoxide metabolites formed. Indeed, a transient intercalation step precedes the formation of DNA adducts [22,23]. This intercalation in the 5 side of the guanine target facilitates further adduct formation by favourably positioning the epoxide for subsequent nucleophilic attack by the guanine N7 [23,24]. The metabolites derived from AFB 2 and AFM 1 could potentially interact with this transient intercalation state, resulting in a decreased mutagenicity of AFB 1 tested in binary or tertiary mixtures. But, since the negative modulation on mutagenicity was also observed without S9mix, additional but as yet unidentified interaction targets must be involved, independently of the metabolic activation step.
Indeed, at least three mechanisms are presumably involved in the mutagenic responses of these aflatoxins: (a) the most powerful and classically described towards TA98 and TA100 upon addition of S9mix involves CYPs in the formation of reactive intermediates such as AFB 1 -exo-8,9-epoxide, and the subsequent binding to the N7 guanine in DNA to form AFB 1 -N7-Gua adducts then, subsequently, the imidazole ring-opened form AFB 1 -FapyGua, this latter being more mutagenic than the former [25,26]; (b) a direct mutagenicity towards these two strains, observed for higher doses of mycotoxins, that did not involve any exogenously activated intermediates; (c) and a mechanism detected in TA102 for the three extracts only upon addition of S9mix, could potentially be subsequent to oxidative damage, since TA102 is responsive to this kind of damage [27].
Verruculogen and fumagillin were produced in the same order of magnitude by the two strains of A. fumigatus that differed regarding their capacity to synthesize gliotoxin. These extracts did not display any direct mutagenicity, but borderline responses were obtained for both extracts in TA102 upon addition of S9mix. When purified standards were tested in parallel, none of them were found mutagenic in the range of concentrations corresponding to the native production of A. fumigatus strains. These data are in accordance with previous results for gliotoxin [11,36]. Even if verruculogen was reported as mutagenic in TA98 and TA100 [14], the tested doses were about four orders of magnitude higher than those tested in the present study. Fumagillin was reported clastogenic, inducing AC and MN [12], but no data were found regarding the mutagenic potency in the Salmonella assay. Thus, as previously suggested for (Flav + ) and (Para + ) extracts, identified mycotoxins or their interactions cannot explain the weak mutagenicity observed in TA102 with S9mix, so additional mutagenic compounds might be produced by A. fumigatus.
To date, interactions between aflatoxins tested in mixtures have not been extensively investigated. Using the Ames test, only Said et al. [37] described a non-additive effect of AFB 1 and N-acetylaminofluorene. Interaction mechanisms were more often studied towards various cell culture models or even in vivo. Hence, recently, Li et al. [31] reported a stronger renal toxicity of AFB 1 and AFM 1 when mixed, involving a pro-oxidant mechanism both in vitro (HEK 293 cells) and in vivo (CD-1 mice). Actually, AFB 1 was most of the time tested in mixture with structurally distinct mycotoxins, such as fumonisin B 1 (FB 1 ) and ochratoxin (OTA). Antagonistic effect of (AFB 1 + FB 1 ) mixture was shown on HepG2 cells cytotoxicity whereas in contrast, synergistic mechanism was obtained on BEAS-2B cell as well as on F-344 rats mortality [38]. Theumer et al. [39] observed a more pronounced apoptosis in rat livers when FB 1 and AFB 1 were mixed, together with enhanced impairments in sphingolipid metabolism. They also reported significant DNA damage in spleen mononuclear cells from rats fed with this mixture [33]. On monkey kidney Vero cells, additive interactions were observed for (AFB 1 + OTA) mix: cell viability was reduced and DNA damage was increased [40]. But in HCT-8 intestinal cancer cells, antagonistic effects were observed on various AFB 1 -induced biomarkers (DNA adducts formation, p53 induction or Mdm2 expression) [41]. Thus, no general rule could be inferred from these data, due to the variety of biological models, protocols, mycotoxin associations and mathematical models used to describe mycotoxin interactions. But the negative interaction observed in the present study for the reconstituted mixtures involving aflatoxins was in accordance with some previous experiments conducted on cell cultures.

Conclusions
After isolation of various fungal strains, we evaluated their toxigenic and mutagenic properties. We thus demonstrated that some of these isolates were able to produce mycotoxins in vitro which could result in a potential hazard for farmers as well as for livestock.
We also showed that, besides identified mycotoxins, additional secondary metabolites should contribute to the global mutagenic responses evaluated with the Ames test.
Among the identified aflatoxins, AFB 1 was the sole mutagenic found when tested from pure standards in a range of doses corresponding to those measured in extracts. However, its association with AFB 2 and AFM 1 led to a decreased mutagenicity, indicating that AFB 2 and/or AFM 1 negatively modulate the toxicity of the corresponding mixture. AFB 1 is rarely produced alone from fungi, and genotoxicity of combined mycotoxins is hard to predict. Thus, mutagenic activity assessment of the "natural" complex mixtures appears more informative and relevant than those of some isolated mycotoxins or reconstituted mixtures for hazard assessment.
Finally, the panel of Salmonella tester strains used indicates that various mechanisms were involved in the mutagenic responses. Besides the well documented formation of AFB 1 8,9-epoxide upon S9mix addition, and the subsequent formation of DNA adducts, a direct mutagenicity was also observed for higher doses in TA98 and TA100 which, therefore, did not involve exogenous activation. An additional contribution through oxidative damage was also detected upon addition of S9mix.

Strain Collection
The non toxigenic A. flavus strain was isolated from a bioaerosol sampled in a dairy cattle shed in Normandy [4]. The toxigenic A. flavus strain was purified from contaminated oilseed cakes, as well as the non toxigenic A. parasiticus strain [2]. The toxigenic A. parasiticus was obtained from a bioaerosol collected during the distribution of cattle feed [3].
The gliotoxin producer strain of A. fumigatus was isolated from a dairy cattle shed bioaerosol [4] whereas the non-producer strain was obtained from a contaminated oilseed cake sample [2].
All isolates were preserved on malt extract agar (MEA) in the laboratory mycological bank (stored at 4 • C).

Toxigenic Ability of Fungal Isolates
Each isolate was tested in triplicate for its ability to produce mycotoxins in vitro. Each strain was cultivated on MEA. The plates were incubated at 25 • C for two weeks and then three agar plugs measuring 8 mm in diameter were removed from the central area of the colony (including conidia and mycelium), pooled, weighted and transferred to 5-mL glass vials as previously described by Garon et al. [42]. Mycotoxins were extracted by 2 mL of ethyl acetate acidified with 1% acetic acid as recommended by Samson et al. [43]. After 15 min of centrifugation at 1500 rpm, each extract was evaporated to dryness under a stream of nitrogen. The residue was dissolved in 0.5 mL of an acetonitrile-water mixture (10:90 v/v) and filtered through Millex HV 0.45 µm before their injection into the high-performance liquid chromatography coupled with mass spectrometry (HPLC-MS/MS). Aflatoxins were then quantified by HPLC-MS/MS in multiple reaction monitoring (MRM) according to the method described by Pottier et al. [44]. Mycotoxins were quantified by HPLC-MS/MS in MRM. Liquid chromatography was performed using an Agilent Technologies 1200 HPLC system coupled to a triple quadrupole spectrometer (6460 series, Agilent Technologies, Santa Clara, CA, USA) equipped with an electrospray interface, operated in the positive and negative modes. The MassHunter B.02.00 software was used for data processing.
The analytes were chromatographed according to 2 chromatographic methods. Eleven mycotoxins (aflatoxins B 1 , B 2 , G 1 , G 2 , M 1 , diacetoxyscirpenol, gliotoxin, mycophenolic acid, neosolaniol, ochratoxin A, T-2 toxin) were separated (Method 1) onto Zorbax SB, Rapid Resolution HT-C18 column (1.7 µm, 50 × 2 mm; Agilent Technologies) at 60 • C. Fumonisin B 1 -13C34 was used as internal standard. The injection volume of the samples on the analytical column was 10 µL. The mobile phase consisted of a variable mixture of acetonitrile (solvent A) and water added with formic acid 1% (solvent B) at a flow rate of 0.4 mL/min. A linear gradient was run starting with 10% to 100% solvent A over 10 min and staying at 100% over 1 min.
The mass spectrometer was operated in positive mode using dynamic MRM. The nebulizer gas and desolvatation gas were respectively nitrogen heated at 300 • C at 10 L/min and 400 • C at 12 L/min.
The mass spectrometer was operated in both negative and positive modes using MRM. Three retention windows were defined according to the retention time and the optimized ESI mode. The nebulizer gas and desolvatation gas were nitrogen heated at 250 • C at 10 L/min (excepted for the third retention window, at 12 L/min) and 400 • C at 12 L/min respectively.
Other common parameters used for the mass spectrometer were as follow: capillary voltage, 4.0 kV; pressure of nebulization, 45 psi; nozzle voltage, 300 V.
The most abundant product ion after collision-induced fragmentation was used for quantitative purposes, and the second product ion for confirmation. The linearity was done by spiking increasing concentrations (triplicate) of the mycotoxin standards (0.1 to 50 µg/L). The quantification and detection limits (QL and DL) were determined by spiked samples based on signal to noise ratio of 10:1 for quantification, and 3:1 for detection limit.

Extract Preparation
For the mutagenic potential evaluation, the dry residues obtained above were dissolved in 550 µL DiMethylSulfOxyde (DMSO) and filtered through Millex PTFE HV 0.45 µm (i.e; resistant Teflon).

Mixture of Mycotoxins
Standard mycotoxins aflatoxin B 1 , B 2 , M 1 (certified grade, Cluzeau Info Labo, Ste Foy la Grande, France); fumagillin, gliotoxin, and verruculogen (Sigma-Aldrich (St. Louis, MO, USA) were tested alone and mixed at identical concentrations to those found in extracts.

Ames Test Procedure
Overnight cultures (12 h at 37 • C with continuous shaking) of Salmonella typhimurium tester strains TA98, TA100 and TA102 (Trinova Biochem, Giessen, Germany) were obtained by addition of 30µL of frozen culture in 10 mL of Oxoid nutrient broth N • 2 solution (0.25 g/10 mL distillated water). The Ames test included a preincubation step as previously described [44]. Briefly, 10 µl of samples (extracts, pure mycotoxins or their mix dissolved in DMSO) were mixed with 100 µl of bacterial overnight culture and either 100 µL of S9mix (prepared with 5% S9) or 100 µL phosphate buffer (for conditions without S9mix). The S9 fraction was obtained from livers of Aroclor 1254-induced Sprague-Dawley male rats (Moltox, Boone, NC, USA). After 60 min shaking (185 rpm at 37 • C), 2 mL of molten agar (agar 0.6% and NaCl 0.5% w/v in distillated water) supplemented with histidine-and biotin traces (final concentration 0.5 g/L and 0.012 g/L for histidine and biotin respectively) were added to the tubes and quickly poured onto minimal glucose plates (containing agar, glucose and mineral salts; for detailed composition, see Maron and Ames [45]). After 48 h incubation at 37 • C, the number of revertants was automatically counted (Noesis software, Saint Aubin, France). For each sample, three concentrations were tested, in triplicate. Results were repeated in two to four independent experiments. Previous studies led us to consider a significant mutagenicity when the ratio (Induced revertants/Spontaneous revertants) was >2 for TA98, ratio > 1.6 for TA100 and ratio > 1.3 for TA102, together with a dose-effect relationship [46]. Toxicity was evaluated in parallel by the microscopic observation of the background lawn density.

Statistical Analysis
An analysis of covariance (ANCOVA) was performed with all significant mutagenic data.
Three kinds of sample were tested, namely fungal extracts, mycotoxin mix and pure mycotoxins. A p value < 0.05 was considered significant. When the difference between these samples was significant, an additional comparison by Student's test using a Bonferroni adjustment was performed.