Introduction

Mycotoxins are secondary metabolites produced by certain mold fungi which, if exposed to humans or animals, can cause serious health consequences (Aichinger et al. 2021; Louro et al. 2024). Exposure to mycotoxins can occur through inhalation or dermal contact but is usually through the consumption of contaminated agricultural products (Barac 2019). Awareness of the risk presented by mycotoxin exposure has increased in tandem with the rising prevalence of not only acute health complications but also the development of a chronic health crisis (Wu et al. 2014; Claeys et al. 2020). As such, efforts have been made to increase regulatory measures and analytical identification methods (Chilaka et al. 2022). Yet, there are still major gaps in mycotoxin research and regulatory policy, highlighted by the need for increased research focused on several specific mycotoxin producing fungal genera, including Alternaria (Aichinger et al. 2021; Qin et al. 2022). The European Food Safety Authority has acknowledged these gaps and in 2016 concluded that exposure to Alternaria produced mycotoxins likely exceeds the threshold of toxicological concern and compound specific analyses of Alternaria mycotoxins should be conducted (Arcella et al. 2016).

Alternaria, generally visible as a black mold, is a major plant pathogen which infects cereal grains, fruit, and oil seeds, among others. Alternaria infection can be especially challenging for agricultural efforts as Alternaria can grow at low temperatures resulting in accumulation of toxins post-harvest in refrigerated conditions (Li et al. 2023). Food infested by Alternaria harbors a host of toxins including alternariol, alternariol monomethyl ether, tenuazonic acid and altertoxin I and II (Puntscher et al. 2018). Several of these toxins have been previously shown to have significant negative health impacts even at low levels. Mutagenic, genotoxic, cytotoxic, (anti-)estrogenic and immunomodulatory effects of isolated toxins and complex extracts have been shown in vitro and in vivo (Aichinger et al. 2019; Puntscher et al. 2019a, b; Gashgari et al. 2019; Hessel-Pras et al. 2019; Schmutz et al. 2019; Hohenbichler et al. 2020).

However, the impact of Alternaria toxin exposure has not been investigated at the metabolomics level in vivo. To fill this gap of knowledge, we performed an untargeted metabolomics analysis of liver samples from rats exposed to Alternaria toxins. This analysis is based on a prior study where the bioavailability, biotransformation, excretion, and toxicity of Alternaria toxins were investigated using a rat model (Puntscher et al. 2019a, b; Aichinger et al. 2022). These initial efforts showed genotoxic effects in vivo of both, isolated ATX-II and an Alternaria-produced toxin mixture designed to simulate the composition of a real-life exposure, although at higher levels than common exposures. Our analysis provides metabolic insights into the biological mechanisms of Alternaria toxin exposure by characterizing the endogenous rat metabolome and screening for Alternaria toxins and other small molecules leading to novel biomarker discovery.

Methods

Chemicals and reagents

LC–MS grade acetonitrile (ACN) and methanol  (MeOH) were obtained from Honeywell. LC–MS grade water was purchased from VWR. Formic acid was purchased from Promochem. Fully 13C-labeled dry yeast extract was procured from ISOtopic solutions (Vienna, Austria) and reconstituted in water per the manufactures' guidelines. Multi-analyte stock solutions of endogenous metabolites and exogenous toxicants (> 100) were prepared containing authentic standards purchased from Sigma and the Toronto Research Company (Supplemental Table 1).

Animal experimental conditions

The in vivo study was conducted as previously described by Puntscher et al. (2019a) and was performed in accordance with the Austrian Animal Welfare Act 2012, BGBI. I Nr. 114/2012(TVG 2012). Briefly, 21 male Sprague DawleyⓇ rats were procured and divided into a control group and two experimental groups. The control group received an innocuous sunflower supplement while the experimental groups received either isolated ATX-II toxin (0.7 mg/kg bw) or a complex extract of Alternaria toxins (50 mg/kg bw) with a matched ATX-II concentration, both in sunflower oil. Animals were harvested via heart punctuation 24 h post exposure at which time organs were immediately put on dry ice and then stored at − 80 °C until liquid chromatography–high-resolution mass spectrometry (LC–HRMS) analysis.

Metabolite extraction

Rat liver samples stored at − 80 °C were allowed to gently thaw at − 20 °C overnight and then on ice until ready for excision. After samples were thawed, ~ 50 mg of material was removed using a scalpel and placed in a vial which was then placed on ice. Samples were kept on ice for the duration of the extraction whenever possible. The collected liver sample sections had a mean weight of 49.8 mg (0.5 mg RSD). Extraction was performed after the addition of 490 µL of MeOH:ACN:H2O (4:4:2). To this mixture 10 µL of isotopically labeled yeast extract and 250 mg of ceramic beads (~ 40 beads) were added after which the vial was placed in a tissue homogenizer (MP FastPreP-24 5G, MP Biomedicals). Using the homogenizer, samples were agitated at 6 m/s for 20 s twelve times with a 5-min rest between each cycle. Samples were then stored at − 20 °C overnight to aid protein precipitation.

After cold storage, samples were placed in a centrifuge at 18,000×g for 15 min. The supernatant was transferred to a new vial and placed in a vacuum concentrator at 4 °C until dryness and stored at − 80 °C until LC–HRMS analysis. Directly prior to analysis, dry samples were reconstituted with 200 µL of H2O:ACN (90:10), agitated via vortex three times for 30 s and sonicated for 5 min. Samples were spun in a centrifuge at 18,000×g for 5 min and finally placed in LC glass vials for analysis.

LC–HRMS analysis

Rat liver extracts were analyzed on a SCIEX 7600 ZENO-TOF mass spectrometer coupled to an Agilent 1290 Infinity II UHPLC. Chromatographic separation was achieved using a Waters Acquity BEH-Amide HILIC column (2.1 mm × 100 mm, 1.7 µm) and a 15-min LC gradient. Water and ACN, both with 0.1% formic acid, were used as mobile phase A and B, respectively. The mobile phase gradient began at 99% B, which was held for 2 min before transitioning to 60% B over the course of 8 min. 60% B was held for 2 min before reverting to 99% B for the final 3 min. The column compartment was kept at 45 °C and the flow rate was kept constant at 0.4 mL/min. Ionization was performed via electrospray ionization in both positive and negative mode in two separate injections. Positive mode ionization was achieved with a spray voltage of 5500 V at 600 °C with a declustering potential at 80 V and a mass range of m/z 100–1000. Negative ionization was completed using a spray voltage of − 4500 V at 600 °C and the same declustering potential and mass range as used in the positive analysis. Fragmentation data were collected from pooled QC samples using the same conditions and the addition of a SWATH-acquisition experiment with mass windows between 50 and 100 m/z. Fragmentation was completed using a stepped collision energy design over the range of m/z 50–1000 which increased from 10 to 55 V with a CE spread of 5–10 V. This fragmentation method allowed for an almost universal precursor fragmentation data capture.

Additional analyses of pooled rat liver extracts, rat liver extracts spiked with the complex mixture, and the complex mixture itself were performed on the same analytical equipment with a different chromatographic column to aid in the annotation of specific less polar analytes with a Waters Acquity HSS-T3 (1.8 µm, 2.1 × 100 mm) reverse-phase column. A 20-min gradient was employed with 0.3 mM ammonium fluoride and ACN as mobiles phase A and B, respectively (Jamnik et al. 2022). The gradient began with 5%B which was held for 1 min before B was increased linearly to 18% at 1.8 min. B was again increased linearly to 35% at 4.2 min, 48% at 13 min and finally 90% at 15.8 min. From 17.7 to 20 min the column was flushed with 98% ACN. The column compartment was held at 40 °C and the flow rate was 0.4 mL/min. Data was collected in both positive and negative mode with SWATH acquisition as previously described.

Data and statistical analysis

MS-DIAL was used to process data including peak picking, data mining, alignment, filtering, and normalization (Tsugawa et al. 2015). Features below an area of 250 were removed and a 0.01 Da window for MS and a 0.05 Da window for MS/MS data were used to align data. Feature datasets were adjusted by accounting for the variation in collected rat liver sample weight and via data-driven locally weighted scatterplot smoothing (LOWESS) normalization accompanied using isotopically labeled internal standards from labeled yeast amino acids spiked into each sample in MS-DIAL [13C-serine for (+) mode and 13C-arginine for (−) mode] to remove systemic sensitivity variation between samples. Fragmentation data collected using SWATH technology was explored using SIRIUS 4 software in combination with CSI:FingerID software and MS-DIAL along with a composite MS-DIAL spectral library for fragmentation matching to annotate features (Dührkop et al. 2015, 2019). An error window of 5 ppm was used for positive annotation along with an 80% spectral library match score. Authentic standard mixtures and 13C—labeled yeast, along with fragmentation data were also used to identify endogenous metabolites and xenobiotics (Supplemental Table 2). Identification levels were assigned to annotations based on confidence criteria proposed by Schymanski et al. (2014). Level 4 annotations relying solely on mass values were not considered while combinations of in silico, spectral library matching, 13C—labeled yeast and authentic reference standards were used to make composite level 1, 2 and 3 annotations. Statistical analysis was completed using MetaboAnalyst and several R packages including ggplot2 and in-house R scripts (Pang et al. 2022).

Quality control

Isotopically labeled yeast were used as a system quality control measure. Fully labeled 13C-yeast were measured before and after positive and negative ionization analytical batches to ensure system reproducibility and stability. Fully labeled amino acids were identified and semi-quantitated. Based on this data coefficients of variation (CV) were determined, which indicated consistency throughout the analytical batch (Supplemental Table 3).

Quality control samples consisting of 100 µL of pooled reconstituted sample extract spiked with 10 µL of 13C yeast were also analyzed before, during and after the analytical batch. In total, five quality control injections were completed. As with the labeled isolated yeast, 13C-amino acids from the labeled yeast spiked into pooled rat liver extract samples were identified, semi-quantified and CV values were assessed. While not as consistent as the matrix-free labeled standards, the CV values still indicate reproducibility throughout the positive and negative ionization analytical batches (Supplemental Table 3). Annotated features were also compared in the pooled samples to ensure reproducibility and metabolites with a CV greater than 30% were removed from the dataset (Supplemental Table 4, Supplemental Fig. 1).

To account for potential background contamination, several blank samples were also generated. Solvent blanks were created using ACN and water taken from the same source as the dilution solvent and mobile phases. An additional extraction blank was created by processing an empty sample in parallel with the liver samples throughout the entirety of the extraction process. The solvent and extraction blank were used to determine systemic and method background noise. Features with abundances five times greater in the pooled rat liver extracts relative to the solvent and processing blanks were retained while features with a ratio under five were removed from the final dataset.

Results

Global untargeted analysis

This study was predicated on the administration of either an isolated mycotoxin, ATX-II, or a complex Alternaria toxin mixture with a profile simulating a real-life exposure scenario. To uncover the metabolic consequences of toxin exposure in our experimental groups, an untargeted analysis was undertaken with data from analytical batches collected in both ESI (+) and ESI (−) modes. The features were then curated and background noise from the analytical system and the extraction procedure were removed. CV values were calculated and features with CV values over 30% were removed. This resulted in 9,310 features in the ESI (+) dataset and 5798 in the ESI (−) dataset. A statistical analysis of metabolomic profiles indicated similar profiles between liver extracts from rats in the experimental groups, illustrated with the use of a 2-D principal component analysis (PCA) (Fig. 1a, Supplemental Fig. 2a). However, a supervised analysis using a 2-D partial least squares discriminating analysis revealed that some separation between the groups existed (Fig. 1b, Supplemental Fig. 2b). Interestingly, more separation was found between the ATX-II group and the control than between the cultured complex mixture and control group. The features providing this separation were elucidated using an ANOVA analysis which revealed 163 statistically significant features in the ESI (+) and 151 statistically significant features in the ESI (−) dataset which were significantly different between the experimental groups (P < 0.05) (Supplemental Table 5).

Fig. 1
figure 1

Untargeted analysis of LC–HRMS data. a 2D PCA of all three experimental groups in ESI (+) mode showing similar profiles indicating a well-balanced dose was applied in the experiment. b 2D PLS-DA of all three experimental groups in ESI (+) separation between the groups that can be elucidated. c Plot of FDR corrected ANOVA P values calculated for identified metabolites between the three experimental groups. Significant metabolites are indicated with red dots. Comparisons denoted with *** have a P value of < 0.01, ** have a P value of < 0.05 and * indicates a P value of < 0.1

Full size image

Statistical analysis of annotated features

An ANOVA analysis of the annotated features elucidated several significantly different features (Supplemental Table 2). Figure 1c shows the annotated metabolites characterized in ESI (+) and (−) mode and their associated ANOVA P value. A total of five annotated metabolites were found to be at significantly different relative concentrations across the groups using a false discovery rate (FDR) corrected ANOVA. These significant metabolites included two acyl-carnitines, butyrylcarnitine and propionylcarnitine, riboflavin, altercrasin A and altersetin, which was isolated in both (+) and (−) mode and found to be significant in both.

Metabolomics analysis of statistically significant metabolites

After isolating statistically important metabolites from the larger dataset, the individual metabolites were examined in closer detail to assess data quality and their statistical relationship between the experimental groups. Beginning with acylcarnitines, data quality was analyzed in the QC samples which demonstrated robust reproducibility (Supplemental Table 4, Supplemental Fig. 1). Although butyrylcarnitine was annotated using in silico methods, an authentic reference standard was used to identify propionylcarnitine, lending confidence to the annotation of both acylcarnitines (Fig. 2a, b, d, e). After ensuring quality, reproducible data, the relative concentrations were analyzed by experimental group (Fig. 2c, f). One of the statistically important acylcarnitines, butyrylcarnitine, was significantly upregulated in the liver extracts from rats receiving the Alternaria complex culture. This trend was calculated to be significant using a t test between the Alternaria complex mixture liver extracts and the ATX-II isolate liver extracts as well as between the Alternaria complex mixture liver extracts and the control liver extracts (P < 0.01). There was no significant difference between the ATX-II isolate and control samples. Propionylcarnitine exhibited similar metabolic regulation and was significantly increased in the Alternaria complex mixture relative to the ATX-II isolate samples. Although samples from rats receiving the Alternaria complex culture indicated an increase when compared to the control samples, the increase was statistically weak (P value 0.068).

Fig. 2
figure 2

Analysis of acylcarnitines isolated in liver extracts. a Isotopic distribution of the acquired data in blue for the m/z value annotated as butyrylcarnitine and the predicted distribution in green. b In silico structural annotation for annotation of butyrylcarnitine using SIRIUS software with predicted structural fragments in purple and predicted formula fragments in green. c Plot of butyrylcarnitine relative concentrations found in each of the three experimental groups. Butyrylcarnitine concentrations were found to be significantly higher in complex toxin exposure group relative to both the ATX-II isolate and control group. d Isotopic distribution of the acquired data in blue for the m/z value annotated as propionylcarnitine and the predicted distribution in green. e Reference standard spectra comparison of propionylcarnitine vs measured spectra in rat livers annotation as propionylcarnitine. f Plot of propionylcarnitine relative concentrations showing a similar relationship as described in a for butyrylcarnitine. Comparisons denoted with *** have a P value of < 0.01, ** have a P value of < 0.05, * have a P value of < 0.1 and n.s. indicates a relationship that was found to be not significant using a t test

Full size image

Another metabolite that was statistically significantly different between experimental groups was riboflavin. Riboflavin was initially annotated by a spectral library match during the untargeted analysis and subsequently identified using a reference standard in a follow-up targeted analysis after purchasing an authentic reference standard. Hence, it represents a confident level 1 identification (Supplemental Fig. 3). Applying a data quality analysis similar to that for the acylcarnitines indicated excellent reproducibility for riboflavin in the QC samples with a calculated CV value of 0.47% (Supplemental Table 4). Riboflavin relative concentrations were investigated between the experimental groups and a strong pattern emerged (Fig. 3a). Riboflavin was calculated to be significantly downregulated in both the Alternaria mixture and the ATX-II isolate groups. There was no significant difference between the Alternaria and ATX-II groups. As riboflavin was found to be associated with similar effects in both mycotoxin-containing groups, it was further examined as a biomarker for Alternaria and ATX-II exposure. A receiver operator characteristic analysis was completed to determine if riboflavin levels could be used to predict both the complex extract exposure as well as ATX-II isolate exposure (Fig. 3b, c). The area under the curve (AUC) values calculated for both analyses were excellent. Riboflavin was slightly more predictive of ATX-II exposure with an AUC of 0.96 relative to the AUC value of 0.88 found for Alternaria culture mixture exposure.

Fig. 3
figure 3

Statistical analysis of riboflavin levels. a Plot of relative concentrations of riboflavin by experimental group in rat liver samples extracts. Riboflavin was significantly downregulated in both the complex mixture exposure group and the ATX-II isolate relative to the control. Comparisons marked by *** have a P value of > 0.01 and comparisons with a n.s. were found to be not significant. b, c AUC measurements for ROC biomarker analyses in comparisons between b control and ATX-II isolate experimental groups and c control and the complex toxin mixture experimental groups. Both biomarker analyses have very high AUCs indicative of low false positive and negative values allowing for high predictive probability making riboflavin an excellent biomarker for Alternaria and ATX-II exposure

Full size image

Metabolic pathway analysis

An analysis of metabolic pathways impacted by the different Alternaria-based exposures was also completed using the full datasets and the pathway analysis function in MetaboAnalyst. A comparison of the ATX-II isolate and the control liver extracts indicated significant differences in riboflavin metabolism (Fig. 4a). Other pathways dysregulated in this comparison included alpha-linoleic acid and retinol metabolism. A comparison of Alternaria extract and control samples again indicated a significant disruption in riboflavin metabolism (Fig. 4b). Several other pathways were found to be dysregulated including glycerophospholipid metabolism, which was found to be significantly dysregulated. Other pathways such as phenylalanine, tyrosine and tryptophan biosynthesis and ubiquinone and other quinone biosynthesis were altered, although they were not as significant. A final pathway analysis of Alternaria complex mixture and ATX-II isolate samples revealed valine, leucine and isoleucine biosynthesis dysregulation as well as moderate effects to retinol metabolism and porphyrin metabolism (Supplemental Fig. 4).

Fig. 4
figure 4

Pathway analysis of identified metabolites. a ATX-II isolate and control groups comparative analysis indicating riboflavin metabolism as a significantly disrupted pathway. b Complex toxin mixture exposure and control groups comparative analysis showing that riboflavin metabolism is significantly disrupted. Significance and pathway impact are plotted on the y- and x-axis, respectively. The size of the dot is indicative of the pathway impact while red colors indicate higher levels of significance. Pathways marked with an * have P values < 0.1, ** denoted pathways have P values < 0.05 and *** indicated pathways have P values > 0.01

Full size image

Alternaria toxin exposomics screening

To determine if Alternaria produced toxins persisted in the liver after 24 h, an investigation into the presence of known Alternaria mycotoxins was conducted. Although concentrations have been previously reported in biological matrices, to our knowledge Alternaria toxins and metabolites have not been explored in liver samples. Commonly reported Alternaria-produced toxins and metabolites were screened for including alternariol, alternariol monomethyl ether, tenuazonic acid ATX-II and ATX-I. Neither ATX-I nor ATX-II was determined in either the isolate or the complex mixture exposed liver extracts, which was consistent with previous results in rat plasma and urine (Puntscher et al. 2019a, b). A further screening for mycotoxins yielded the presence of altersetin, a toxin with antibiotic properties produced by Alternaria, which was found exclusively in the rats receiving the complex mixture (Fig. 5a, b) (Hellwig et al. 2002). Altersetin was annotated using both spectral library matching and in silico predictive modeling and had a CV of 2.5% in pooled QC samples, indicating low technical variability (Supplemental Fig. 5). An analysis of pooled samples from rats receiving the complex mixture and the mixture itself revealed the presence of altersetin in the complex mixture with the same m/z and retention time found in the rat liver extracts (Fig. 5c). Additional reversed-phase LC–HRMS analysis, found that altersetin had a retention time of 16 min and showed the same pattern with respect to pooled rat samples and the complex mixture as with the initial HILIC analytical analysis (Fig. 5d).

Fig. 5
figure 5

Alternaria-specific metabolite levels. a Relative concentrations of altersetin by experimental group, isolated solely in livers from rats exposed to the complex toxin mixture (“Extract”). b–d EICs of m/z 400.247, the value associated with altersetin, for b individual rat liver extracts from rats exposed to the complex mixture analyzed using the HILIC method, c pooled liver extracts from rats exposed to the complex mixture and the complex mixture analyzed using the HILIC method and d pooled rat liver extracts from rats receiving the complex mixture and the complex mixture analyzed using the reverse-phase method. e Relative concentrations of altercrasin A by experimental group, isolated solely in livers from rats exposed to the complex toxin mixture (“Extract”). f–h EICs of m/z 414.226, the value associated with altercrasin A, for f individual liver extracts from rats receiving the complex mixture analyzed using the HILIC method, g pooled liver extracts from rats exposed to the complex mixture and the complex mixture analyzed using the HILIC method and h pooled liver extracts from rats receiving the complex mixture and the complex mixture analyzed using the reverse-phase method

Full size image

Although most common Alternaria produced toxins were screened for and not detected in liver samples after 24 h, a deep exploration of the data via non-targeted analysis revealed the potential presence of the toxin altercrasin A, which was isolated using an in silico analysis and only found in the rats receiving the complex mixture (Fig. 5e, f). Since neither an authentic reference standard nor an MS/MS library entry is available for this toxin to the best of our knowledge, the annotation is assigned a confidence level of 3 (Supplemental Fig. 6).

To further investigate this potentially interesting finding, the complex mixture from the cultured extract was examined independently using the same LC–HRMS method and the m/z associated with altercrasin A was found at the same retention time as in the rat liver extracts (Fig. 5g). This demonstrates that the annotated feature likely originates from the extract. To investigate this peak of interest further, a sample of the liver extract and the cultured extract were subjected to complementary reversed phase LC–HRMS and, as with the HILIC analysis, the m/z for altercrasin A and retention times between the sample types matched in both, the rat liver extracts and the complex mixture administered during the experiment (Fig. 5h). The analysis of altercrasin A revealed good peak shape in an extracted ion chromatogram of the MS1 data and MS/MS collected data indicated a matching isotopic distribution as well as promising structural fragmentation matching annotation (Supplemental Fig. 6a, b). Technical variability in the pooled QC samples was examined and a CV value of 12% was calculated for the m/z value associated with altercrasin A (Supplemental Table 4).

Discussion

This study investigated the in vivo metabolic effects of Alternaria toxin exposure in a rat model, building on analysis from the same animal experiments which elucidated important insights into the absorption, metabolism and excretion of Alternaria toxins, especially ATX-II (Puntscher et al. 2019a, b). Toxins were found to be greatly reduced in concentration after three hours while some toxins were still isolated at very low levels in some biological matrices after 24 h. Additionally, the genotoxicity effects of Alternaria, and specifically ATX-II, were described in vivo (Aichinger et al. 2022). Genotoxicity was determined in the colon of the rats receiving an Alternaria complex mixture and those receiving an ATX-II isolate. By comparison, an analysis of liver samples showed little genotoxic effects. Surprisingly, the ATX-II isolate demonstrated a higher genotoxic effect in the colon relative to the mycotoxin mixture, indicating a potential source of mycotoxin metabolism or protection activation provided by components of the complex mixture as it contained the same concentration of ATX-II. Although these studies expanded the understanding of Alternaria exposures, gaps remained. By harnessing LC–HRMS technology, we were able to investigate the effects of exposure in vivo and determine dysregulated metabolites and metabolic pathways as well as isolate a biomarker for Alternaria exposure. Our analysis indicated a complex metabolic shift associated with Alternaria exposure highlighted by the persistent presence of the Alternaria produced mycotoxin altercrasin A and metabolite altersetin isolated from Alternaria exposed rat livers 24 h post-exposure as well as an increase in acyl-carnitines in rats exposed to an Alternaria mycotoxin mixture and the downregulation of riboflavin levels in livers from rats exposed to either the Alternaria complex mixture or an isolated Alternaria mycotoxin, ATX-II.

An interesting component to earlier studies was the finding that ATX-II exposed rat colons exhibited higher genotoxic damage markers than the complex mixture (Aichinger et al. 2022). Although genotoxic markers were not seen in the liver, a similar result was found with respect to the global untargeted data analysis. A supervised approach to determine the variation between each experimental group indicated that the liver samples from rats exposed to the Alternaria mycotoxin mixture were more similar to the control than the ATX-II isolate (Fig. 1b, Supplemental Fig. 2b). These results, taken with conclusions reached concerning the toxicity protection offered in the rat colon from the complex mixture relative to the ATX-II isolate demonstration in the previous study, seem to point to a mixture effect in the Alternaria cultured complex mixture. This could be, as previously conjectured, an increase in the production or release of a metabolite with a protective or mediating function or the activation of an enzyme capable of metabolizing toxins produced by Alternaria.

Although liver samples did not show genotoxic effects before, our metabolomics analysis indicated several differences in liver samples 24-h post-exposure. An important finding is that of the toxins altersetin and altercrasin A in the liver samples from rats exposed to the Alternaria mycotoxin mixture (Fig. 5a and e). Little is known about altercrasin A, a decalin derivative produced by Alternaria, yet its persistent presence in the liver is notable and further research should be completed (Yamada et al. 2019; Jiang et al. 2020). As mycotoxins have been shown to negatively impact liver metabolism and health, this could be an important component for acute and chronic Alternaria exposure risks (Hasuda et al. 2022).

Along with altercrasin A, short chain acyl-carnitines were calculated via a statistical analysis to be upregulated in livers from rats exposed to the Alternaria cultured mixture. Specifically, the acylcarnitines butyrylcarnitine and propionylcarnitine demonstrated elevated levels in the complex mixture samples relative to the control and ATX-II isolate samples (Fig. 2c, f). Acylcarnitines are esters of L-carnitine and fatty acids which transport fatty acids across the mitochondrial membrane and initiate the β-oxidation pathway, making them integral to energy metabolism. Acylcarnitines are, therefore, vital to energy metabolism and are often used as biomarkers for mitochondrial activity and to screen for disorders effecting β-oxidation (McCann et al. 2021). Recently, the use of acyl-carnitines as markers for disease states has expanded as acylcarnitines have been posited as biomarkers for a wide range of diseases including diabetes, cancers and cardiovascular disease (Dambrova et al. 2022).

The liver, specifically, has been shown to be a major systemic contributor of not just acylcarnitines, but short chain acylcarnitines such as butyrylcarnitine (Xu et al. 2016). Elevated levels of short chain acyl-carnitines have also been reported to be associated with increased inflammation and disease states in the liver (Ghosh et al. 2016; Li et al. 2019). Taken together, it is likely that a component of the complex mycotoxin mixture is still impacting liver energy metabolism 24 h after exposure. Given the results from our Alternaria compound screening analysis focused on mycotoxins, this increase in acylcarnitines could be due to effects from a persistent component of the complex toxin mixture.

Riboflavin was also shown, like acylcarnitines, to be differentially regulated in our experimental groups through our metabolomics analysis (Fig. 3a). Alternaria exposure, either via the cultured Alternaria or the ATX-II isolate exposure led to significant downregulation in riboflavin concentrations. The degree of downregulation was similar between both routes of exposure indicating ATX-II alone impacts riboflavin metabolism. The variation in riboflavin levels between the groups receiving either of the Alternaria exposures was statistically robust enough that riboflavin was found to have strong statistical value as a biomarker and predictor of exposure (Fig. 3b, c).

Riboflavin, or vitamin B2, is an essential nutrient which plays vital roles in several metabolic pathways such as central carbon energy metabolism as the precursor for both flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) and stress responses including exhibiting anti-inflammatory effects (Ashoori and Saedisomeolia 2014; Barile et al. 2016; Ahn and Lee 2020). Human and animal model studies have demonstrated the negative impacts of riboflavin deficiency which include growth retardation, fatigue, depression, lipid accumulation and nutrient absorption (Thakur et al. 2017; Mosegaard et al. 2020). For example, decreased riboflavin concentrations have been shown to limit iron uptake in the small intestine leading to anemia in human populations (Aljaadi et al. 2022). Liver specific pathologies have also been attributed to riboflavin deficiency including the development of fatty liver and liver enlargement. Tang et al. (2022) demonstrated that certain liver pathologies were regulated by specific proteins which were associated with riboflavin deficiency. Proteins integral to fatty acid β-oxidation and mitochondrial electron transport chain function were decreased while triacylglycerol and cholesterol biosynthesis were increased. Ultimately, these results provide a mechanistic avenue for the development of liver metabolic disorders arising from riboflavin deficiency.

Riboflavin deficiency can be problematic as riboflavin is not endogenously produced in animals and humans and is therefore recovered solely from dietary sources such as dairy and meat products, where it is prevalent (Suwannasom et al. 2020). Since riboflavin needs to be acquired via dietary measures, absorption rates in the intestinal tract will greatly influence systemic and tissue specific levels (Levy and Jusko 1966). Riboflavin is transported from the gut by a group of solute carrier transporters (SLC) type transporters. Characterized riboflavin transporters include RFVT1, RFVT2 and RFVT3 and are integral to the absorption of riboflavin, which is thought to occur mostly in the small intestine where RFVT1 is expressed on the apical side and RFVT2 and RFVT3 are expressed on the basal side (Yonezawa and Inui 2013; Udhayabanu et al. 2017) (Fig. 6). It is likely, given our data, that toxins produced by Alternaria alter this pathway at some point leading to a decrease in riboflavin uptake across the epithelial cell layer in the intestinal tract. This has been demonstrated previously with the mycotoxin deoxynivalenol (DON) which was shown to limit uptake of glucose, 5-methyltetrafolic acid, molybdenum, and magnesium in the intestinal tract (Pinton and Oswald 2014). These results are also in line with the conclusions from previous Alternaria toxin exposure results in rat models where the gastrointestinal tract was impacted by both the ATX-II isolate and a complex Alternaria toxin mixture (Aichinger et al. 2022).

Fig. 6
figure 6

Riboflavin metabolism and Alternaria toxin action. The journey of riboflavin begins with absorption in the gut followed by transport to the liver and, ultimately, the mitochondria. The role of riboflavin in β-oxidation is also shown as well as integrated data collected from our study indicating riboflavin and acylcarnitine level disparities between experimental groups. Red X marks proposed disruptions in the pathway arising from Alternaria toxin exposure and include limited riboflavin uptake as well as β-oxidation suppression stemming from inhibition of Acyl-CoA Dehydrogenase due to a lack of essential FAD

Full size image

Once across the intestinal epithelium, riboflavin carrier proteins (RCP) transport riboflavin via blood to targeted areas of the body, including the liver where small amounts of riboflavin are stored. Riboflavin shuttled to the liver enters hepatocytes through RFVT2 transporters and is then transported across the mitochondrial membrane through an unknown mechanism (Yao et al. 2013). Once in the mitochondrial matrix, riboflavin is catabolized by riboflavin kinase to form FMN. FMN can be further modified by the activity of FAD synthetase to generate FAD, a required cofactor in β-oxidation.

β-Oxidation, or the catabolism of fatty acids, only occurs in the mitochondria which presents challenges due to the negative charges associated with fatty acids. This makes crossing the mitochondrial membrane difficult and requires a transporter and modifications to make possible. The journey of fatty acids to the mitochondria begins in the cytosol, where fatty acyl-CoA synthetase forms a thioester bond between fatty acids and coenzyme A, creating an acyl-CoA (Watkins 2013). Following the synthesis of acyl-CoA, carnitine acetyltransferase activates transfer of the acyl group to carnitine to create acylcarnitines. Acylcarnitines are then shuttles across the membrane via the carnitine acylcarnitine carrier (CAC) (Tonazzi et al. 2013) (Fig. 6). Once in the mitochondrial matrix, the carnitine is removed and replaced by coenzyme A. The resulting fatty acyl-CoA then enters the β-oxidation cycle. Each cycle of β-oxidation cleaves two carbons, released as acetyl-CoA and two electron carriers, NADH and FADH2. The first step in the process requires the reduction of FAD to FADH2 and without FAD, β-oxidation will not occur (Swigoňová et al. 2009). In this way, FAD regulates β-oxidation and the conversion of acylcarnitines to acyl-CoA (Adeva-Andany et al. 2019). Since FAD production in hepatocytes is influenced by riboflavin transport to the liver and across the intestinal epithelium, decreased riboflavin uptake would lead to a decrease in β-oxidation and an increase in acyl-carnitines. This result is seen in the rat livers exposed to the complex toxin mixture but not in rat livers exposed to solely ATX-II. A component in the Alternaria complex mixture, possibly altercrasin A, appears to be decreasing FAD concentrations in the mitochondria, leading to the inability to start the conversion of acyl-carnitines in β-oxidation. This could be accomplished by limiting riboflavin transport into the mitochondrial matrix or by inhibiting riboflavin conversion to FMN or FAD or by disrupting β-oxidation itself, compounding the initial decrease in riboflavin concentrations likely due to decreased intestinal absorption.

Alternaria and their associated toxins represent an underestimated health risk which has been compounded by a lack of information concerning toxin metabolism and their systematic investigation of effects on the metabolome. Previous research has shown Alternaria to be a potent toxin producer with genotoxic activity. Here we present additional findings indicating that Alternaria exposure results in an accumulation of short chain acylcarnitines in the liver as well as likely leading to a decrease in riboflavin absorption in the intestinal tract translating to a decrease in liver concentrations. These physiological changes would limit β-oxidation resulting in decreased energy production while at the same time reducing the availability of a functional vitamin like riboflavin and the associated coenzymes FAD and FMN, causing negative health consequences. These results, taken with previous work indicating the genotoxicity of certain Alternaria toxins, indicate the importance of further health-based investigations and the potential need for regulations regarding Alternaria toxins in food and animal feed.