Aflatoxins are natural scavengers of reactive oxygen species

The role of aflatoxins (AFs) in the biology of producing strains, Aspergillus sect. Flavi, is still a matter of debate. Over recent years, research has pointed to how environmental factors altering the redox balance in the fungal cell can switch on the synthesis of AF. Notably, it has been known for decades that oxidants promote AF synthesis. More recent evidence has indicated that AF synthesis is controlled at the transcriptional level: reactive species that accumulate in fungal cells in the stationary growth phase modulate the expression of aflR, the main regulator of AF synthesis—through the oxidative stress related transcription factor AP-1. Thus, AFs are largely synthesized and secreted when (i) the fungus has exploited most nutritional resources; (ii) the hyphal density is high; and (iii) reactive species are abundant in the environment. In this study, we show that AFs efficiently scavenge peroxides and extend the lifespan of E. coli grown under oxidative stress conditions. We hypothesize a novel role for AF as an antioxidant and suggest its biological purpose is to extend the lifespan of AFs-producing strains of Aspergillus sect. Flavi under highly oxidizing conditions such as when substrate resources are depleted, or within a host.

www.nature.com/scientificreports/ thesized only AF congeners B 1 and G 1 ; notably, the amount of AFG 1 was significantly higher than that of AFB 1 (87.1, 102, 123, 132 µg/400 mL). This difference is possibly related to oxygen consumption and carbon dioxide accumulation (Fig. 1C. In the CS, O 2 concentration decreased from 20 to 5% at 72 h, leading to hypoxic conditions for A. parasiticus as early as 96 h. As expected, the trend of CO 2 concentration constantly increased up to 45% at 168 h of fungal growth, in the CS. In the OS, O 2 and CO 2 starting parameters (O 2 = 20.58% and CO 2 = 0.085%) stayed constant throughout growth (data not shown).
Glucose uptake registered a very similar trend in both systems (Fig. 1D), which implies a comparable metabolic energy expenditure.
Aflatoxins demonstrate antioxidant capacity against abiotic oxidants. According to current literature and previous results, oxygen is deemed to control AF synthesis 14,15 . In an OS, A. parasiticus accumulates reactive species (RS) during growth, and especially during the late log/stationary phase 16 . RS are potentially toxic to the fungus itself. To counteract their effect, Aspergillus secretes several antioxidant compounds 17 . In our experimental setting, we found that a substantial amount of AFs (up to hundreds of ppb) was secreted during the late log/stationary phase. For this reason, we hypothesized that possibly AFs are countering oxidants, allowing and/or extending fungal survival during the stationary phase.
To explore this possibility, we assayed the potential of the four AFs (B 1 , B 2 , G 1 and G 2 ) to scavenge RS produced by the oxidant 2,2'-Azobis, 2-amidinopropane (APAB) (see methods for details). We carried this out in hydrophilic and lipophilic environments via crocin bleaching test (Fig. 2). The result showed that, among the four AFs, the antioxidant capacity was as follows: G 1 > B 2 > G 2 > B 1 (AF produced by our strain in bold) in the hydrophilic environment. There was no or minimal antioxidant effect in the lipophilic environment. In addition, AFG 1 presented an antioxidant value (K a /K c = 2.49) comparable with that of the hydrophilic fraction of some polyphenols known for their important antioxidant activity [18][19][20] .
To act effectively as an antioxidant, in presence of an oxidant AF should transform to a stable antioxidant compound 21 . Therefore, we analyzed the reaction products for the four AFs exposed to APAB oxidation. In every reaction, we recovered the dihydrodiol byproduct of AF (Fig. 3).
In Fig. 4, we postulated the reaction scheme for AFB 1 , the most toxic of AFs. Based on known AF oxidation products, the peaks identified from the mass spectra seemed to be in agreement with the postulated reaction scheme outlined above 22 .
Although we did not see the epoxide peak in this data set, it seems reasonable for the dihydrodiol peak formation to go via epoxide. There were good mass accuracies for the AFB 1 dihydrodiol peaks, [M + H] + m/z 347.0766 with − 0.26 ppm error, [M + Na] + m/z 369.058 with − 0.56 ppm error, with good mass spectral isotope matches. There was a number of peaks showing the mass m/z of 347, but this could be due to the formation of the dihydrodiol from the endo, exo-8,9-epoxides, resulting in the hydroxide in the different configurations; one peak may be attributed to the formation of the AFB 1 dialdehyde.  www.nature.com/scientificreports/ Aflatoxins exhibit antioxidant capacity for cultures of a model organism Escherichia coli K-12. E. coli K12 was exposed to hydrogen peroxide in a 0 to 0.8 mM range. Growth rate was severely affected up to 0.45 mM, and less so beyond that threshold (Fig. 5A). Indeed, OD values, monitored to estimate cell density, increase up to 0.1 mM H 2 O 2 and decrease at higher concentrations (Fig. 5B). Intriguingly, the analysis showed an increase in maximum OD (Fig. 5B) for all the cultures exposed to H 2 O 2 in the presence of 20 µg/mL AFB 1 , compared to the control. This overall advantage in population size is not a consequence of increased growth rate which, conversely, registers lower values than the control. Of notice is the reduced variability across the sample readouts of the toxin-supplemented instances: the toxin exerts a stabilizing effect on the overall cellular growth rate in presence of the oxidative stress, whereas values of growth rate and OD show more variability when the toxin is not present, as highlighted by the error bars in the graph (Fig. 5A

Discussion
Sources of oxidants are everywhere around natural organisms: light, UV radiation, oxygen, and metals can be responsible for the formation of "environmental" RS. Eventually, oxidative stress is a condition all cells have to face to live in the presence of oxygen (and in particular singlet oxygen, 1 O 2 ) 23 . In fact, RS are, to some extent,  www.nature.com/scientificreports/ a normal by-product of cell metabolism 24,25 . Aflatoxigenic fungi apparently have found a special way to better survive and/or prolong their survival under oxidative stress conditions. To assess the relation between AF synthesis and oxygen content, we set up experiments under two growth conditions: a closed system (CS) and an open system (OS). In the CS, constant depletion of oxygen and accumulation of CO 2 do not support AF synthesis. In the CS, A. parasiticus grows similarly to the OS condition, even if, as expected, in the CS the amount of mycelium is significantly lower than in the OS. Nevertheless, glucose consumption showed similar trends in the OS and CS conditions. From these observations, we infer that oxygen is necessary for AF synthesis, that is, oxygen is a conditio sine qua non for AF production, independent of mycelial growth. Hypoxia is reported 26 to lower metabolic rates in toxigenic Aspergilli; this could explain the reduced fungal growth in the CS compared to the OS. Furthermore, the CS might drive A. parasiticus to hypoxia and therefore to a pro-glycolytic phenotype (similar to cancer's Warburg effect) exploiting glucose with a reduced energetic yield. Under these conditions, RS should be present in lower amount as indicated in A. flavus 27 and mammalian cells. Other studies demonstrated the key role played by endogenous 4,14,15 and exogenous 10,28-30 oxidative stress in AF biosynthesis. These observations point to oxidative stress and oxygen as key players for AF production in Aspergillus sect. Flavi. An aflatoxigenic fungus can survive in a highly oxidant environment, as demonstrated by studies in which culture media were supplemented with lipophilic epoxides 31 or by hydroperoxides of linoleic and linolenic acids 29 . AF production in this case is highly stimulated in correlation to RS concentration 29 : the higher the RS concentration, the higher the AF production. This fungal behavior begs to inquire if there is a link between fungal survival in a toxic environment and AF production. Some authors studying the metabolism of AF producing and non-producing fungi, suggest that the formation of AF may occur as a "compensatory" response to ROS accumulation 15 . In relation to these aspects, we hypothesized an antioxidant role per se of AF and measured their antioxidant capacity. In this assay (crocin test; see Methods), the hydroperoxides generated by APAB should react with AF in lieu of crocin, if AF displayed remarkable antioxidant features 32,33 . Indeed, AFs showed, to different extents (G1 > B2 > G2 > B1), significant antioxidant capacity, comparable to other synthetic and natural antioxidants molecules [21][22][23] . Notwithstanding their stability under radical attack, a continuous source of ROS, such as ROO· originated by APAB, can (di)hydroxylate the furanose of AFs, as suggested by mass spectrometer results. This reaction allows the molecule to acquire two hydrophilic groups (-OH), in a hydrophilic environment. This might partly explain the antioxidant capacity of AFs. This aspect requires additional work to be verified.
Why should AFs act as antioxidants? What is the benefit for the fungus? Results on E. coli suggest that AF might enhance the resilience of cells to oxidative stress, prolonging their lifespan. Previous in vitro and in vivo studies indicate that A. parasiticus can endure high and prolonged oxidative stress conditions, during the late log/stationary phase in a culture medium 16 as well as on stored plant seeds enriched in peroxides 32 . Recently, Linz group elegantly demonstrated that in A. parasiticus AFs are under the control of the oxidative stress-related factor AP-1-like 33 . In this model, A. parasiticus uses the canonical scavengers for RS (e.g. catalases) during the lag to early log phase of growth and AP-1 starts transcribing AflR only in the late log/stationary phase of growth, switching AF synthesis on. Within our study, it emerges that AFs are produced, and can therefore scavenge oxidants within a specific time-range: in the late log/stationary phase in which the natural antioxidant capacity of the organism (such as catalases) are probably less-active. Within this frame, evidence points at AF acting as an antioxidant to allow the fungus to extend its survival in highly oxidizing environments playing a role as another fungal adaptive strategy for life.

Materials and methods
Experimental design. The experiments were carried out in two different systems of incubation (Fig. 7).
An open system (OS) in which the Erlenmeyer flask was closed by cotton and a free exchange of air between the system and the environment was established. In the closed system (CS) an airflow was pushed into the Erlenmayer flask contain 20.58% of O 2 and 0.08% of CO 2 . The flask was then hermetically closed in order to prevent air exchange between the gases produced inside and the external environment. This closed system (CS) was connected to an O 2 /CO 2 COMBY CHECK analyzer (Dansensor Italia s.r.l., Italy) to measure the O 2 /CO 2 ratio inside the flask during fungal growth. Culture conditions. Aspergillus parasiticus NRRL 2999, an aflatoxin producer, was used in these experiments. The fungal isolate was grown at 28 °C on Czapek Dox Agar (CDA) for 7 days prior to inoculation. The inoculum was performed by suspending conidia of A. parasiticus in 2 ml of sterile H 2 O + Triton X100 (0.1% w/v) and inoculating 1 × 10 6 conidia in 400 mL of glucose 3% w/v and peptone 1% w/v broth (PDB). The cultures were incubated at 28 °C for 24, 48, 72, 96, and 168 h. Every 24 h, the cultures were filtered by Millipore filter (pore size 0.45 µm) and washed three times in a saline solution (NaCl 0.9% w/v). The mycelium was dried at 80 °C for 48 h and then weighed. The consumption of glucose has been estimated with Boehringer Mannheim (Germany) D-glucose kit.
Determination of hydrophilic and lipophilic antioxidant capacity. The antioxidant capacity of each single AFs has been estimated by crocin bleaching inhibition method 36 This method is based on the bleaching of crocin as a result of its oxidation by a source of radicals, APAB ([2,2'-Azobis(2-methylpropionamidine) dihydrochloride] and AMVN (2,2'-azobis (2,4-dimethylvaleronitrile). The reaction is monitored by recording, for ten minutes, the corresponding decrease of absorbance at 443 nm. The reaction with crocin alone gives us the bleaching rate V 0; when an antioxidant or pseudo-antioxidant is added, it reacts with the free radicals and, as a consequence, crocin bleaching rate (V a ) is reduced, according to the competitive reaction equation: Aflatoxin oxidation products characterization. For the discovery of AF degradation products, four solutions containing different aflatoxins (AF B 1 , B 2 , G 1 or G 2 ; 0.2 mg each) and APAB (4 mg) have been dissolved in 4 mL of water and kept at 40 °C for 30 min. Then, AFs have been extracted from 1 mL aliquots of the solution (as reported below) for analysis. LC/MS analysis was performed with the following equipment and reagents: Kinetex column 2.6 µM EVO C18, 100 × 2.1 mm; mobile phase A: Water 5 mM Ammonium Acetate, 0.5% Acetic Acid; mobile phase B: Methanol, 5 mM Ammonium Acetate, 0.5% Acetic Acid, at a flow rate of 350 µl/min, and with UV wavelengths at 354 and 360 nm ( Table 1). The eluent from the column was directed into the electrospray source of an Agilent 6220 TOF mass spectrometer operated in positive ionization mode. Data was converted into mzML file format and analyzed using MZMine software.
In vitro test of aflatoxins antioxidant capacity. Escherichia coli K12 cell cultures were exposed to different concentrations of hydrogen peroxide (H 2 O 2, Thermo-Fisher Scientific, Waltham, MA, USA) in the presence of AFB 1 . E. coli was selected for the purpose of this experiment because it lacks the enzyme Cytochrome P450, whose interaction with AFB 1 is the mechanism through which the toxin causes its deleterious effect to cells 37,38 . Therefore, in this experimental context, the antioxidant effect of AF does not come with a viability cost for the bacterial cells. Defined medium "Z" (KH 2 PO 4 1.5 g/L; K 2 HPO 4 × 3H 2 O 3.8 g/L; (NH 4 ) 2 SO 4 1.3 g/L; Na citrate × 2H 2 O 3.0 g/L; Glucose 4.0 g/L; 1 M MgCl 2 ; 1 M CaCl 2 , pH 7.2) was used to grow E.coli K12 strain MG1655 at 37 °C in agitation. AFB 1 (Cayman Chemical Company, Ann Arbor, MI, USA) was dissolved in methanol and added to the culture medium at the concentration of 20 µg/mL. H 2 O 2 was added to the culture medium at thirteen different concentrations from 0.1 to 0.7 mM (0.05 increments). Control samples were prepared by adding the respective amount of methanol without the toxin of the AFB 1 -supplemented cultures, and loaded on the same multiwell plate. Each combination of treated and control samples counted ten replicates for all the different concentrations of peroxide tested. For the purpose of data analysis, the most peripheral wells were excluded to rule out evaporation-related variability. The cell viability test was run on a 384 multiwell plate on a Synergy™ Mx Multi-Mode Microplate Reader (Agilent Technologies, Santa Clara, CA, USA). Cellular growth was assayed using absorbance readings at 600 nm (OD 600 ). AFB 1 concentration was monitored via fluorescence assay (ex. 380 nm, em. 480 nm). Experimental time was set at 24 h; absorbance and fluorescence measurements were obtained every 4 min for a total of 361 reads. Growth rate was defined as the rate of exponential increase in OD values in the range between 0.05 and 0.2 and calculated in Matlab using polyfit. A second experiment with E. coli K12 cultures was performed including alternatively AFB 1 (Cayman Chemical Company, Ann Arbor, MI, www.nature.com/scientificreports/ USA) and PhR (Thermo-Fisher Scientific, Waltham, MA, USA) as antioxidants. The cell viability test was run on a 96 multiwell plate on a Synergy™ H4 Multi-Mode Microplate Reader (Agilent Technologies, Santa Clara, CA, USA). Cellular growth was assayed using absorbance readings at 600 nm (OD 600 ). Experimental time was set at 24 h; absorbance measurements were obtained every 5 min for a total of 289 reads. H 2 O 2 was added to the culture medium at five different concentrations from 0 to 0.4 mM (0.1 increments). Growth rate was defined as the rate of exponential increase in OD values in the range between 0.05 and 0.2 and calculated in Matlab using (fit_logistic.m); edge wells, potentially affected by evaporation, were discarded from the analysis.