Aflatoxin B1 (AFB1), which is a mycotoxin produced mainly by Aspergillus flavus and Aspergillus parasiticus, occurs naturally in human foods and animal feeds (Zhou et al. Reference Zhou, Xie, Zhang, Stewart, Gu and Ryan2006). Acute or chronic aflatoxicosis, hepatotoxicity, teratogenicity, and immunotoxicity in humans and animals as a consequence of AFB1 exposure has been reported by many researchers (Sabbioni & Sepai, Reference Sabbioni, Sepai, Sinha and Bhatnagar1998; Smela et al. Reference Smela, Currier, Bailey and Essigmann2001). It has been well documented that AFB1 reduces performance, alters blood profiles, decreases immune function, and influences antioxidant status as well as liver function in animals (Bennett & Klich, Reference Bennett and Klich2003). For this reason, AFB1 is classified as a group 1 carcinogen in humans by the International Agency for Research on Cancer (IARC), which is part of the World Health Organization (IARC, 2002). In mammals, AFB1 is metabolised to AFM1, which also has carcinogenic potency (Creppy, Reference Creppy2002) and can appear in milk. Therefore, AFM1 has been classified as a probable human carcinogen (IARC, 2002). The European Commission has established the maximum admissible levels of AFB1 as 20 µg/kg in dairy animal feeds, and the allowable level of AFM1 in milk and milk products should not exceed 50 ng/kg (Commission, Reference Commission2006). In China, AFB1 and AFM1 are also regulated by the Standardisation Administration of China at 20 µg/kg (SAC, 2001) and 50 ng/kg (SAC, 2003), respectively.
It is widely known that AFB1 is biotransformed by the hepatic microsomal cytochrome P450 in the liver into reactive AFB1-epoxide (AFBO). AFBO binds strongly to proteins and imparts its effect, which alters the normal biochemical functions of protein macromolecules, which can lead to deleterious effects at the cellular level (Mishra & Das, Reference Mishra and Das2003). AFBO binds strongly to DNA and RNA as well, causing aflatoxins to impart their effect by means of interfering with DNA replication and transcription of messenger RNA into protein at the molecular level. Mutation is caused by binding of the aflatoxin molecule to DNA and subsequent erroneous protein synthesis (Bedard & Massey, Reference Bedard and Massey2006). Therefore, AFB1 is a biosynthetic inhibitor both in vivo and in vitro, which not only affects different metabolic pathways, such as glycogenolysis/glycolysis (Kiessling & Adam, Reference Kiessling and Adam1986) and phospholipidation (Hsieh et al. Reference Hsieh, Hsu, Chen and Santella1988), but also changes amino acid transportation (Mclean & Dutton, Reference Mclean and Dutton1995). Smith & Moss (Reference Smith and Moss1985) reported that aflatoxin reduces the hepatic glycogen level. AFB1 produced various alterations in total lipid and cholesterol as well as the fatty acid composition of hepatic phosphatidylcholine and phosphatidylethanolamine in rats three weeks post-exposure (Baldwin & Parker, Reference Baldwin and Parker1986).
Metabonomics is a powerful approach to delineating the global metabolic profiling of living organisms and capturing the subtle metabolic alterations associated with physiological and pathological stimulations. 1H Nuclear magnetic resonance (NMR)-based metabonomics, with its simplicity of sample preparation and robust reproducibility, has been extensively used in metabonomic analysis of biofluids and tissues to study metabolic changes. The study of the endogenous metabolic alterations induced by heavy metals was used in NMR-based metabonomics, with results showing that the changes in lipid fraction, unsaturated lipids, and in the level of amino acids suggest perturbation of the metabolism of lipids and amino acids (Dudka et al. Reference Dudka, Kossowska, Senhadri, Latajka, Hajek, Andrzejak, Antonowicz-Juchniewicz and Gancarz2014). Sundekilde & Poulsen (Reference Sundekilde and Poulsen2013) identified the biomarkers of somatic cell counts by NMR-based metabonomics including lactate, acetate, isoleucine, hippurate, butyrate, and fumarate. NMR-based metabonomics has also been applied to the toxicology investigation of the T-2 toxin (Wan et al. Reference Wan, He, Deng, Hao, Tang and Wang2016), ocharatoxin A (Sieber et al. Reference Sieber, Wagner, Rached, Amberg, Mally and Dekant2009), deoxynivalenol (Hopton et al. Reference Hopton, Turner, Burley, Turner and Fisher2010), and streptozotocin (Diao et al. Reference Diao, Zhao, Guan, Zheng, Chen, Yang, Lin, Chen and Gao2014).
For the abovementioned reasons, we conducted this study to analyse the metabolic alterations caused by AFB1 in dairy goat serum using NMR-based metabonomics approaches. The objectives of this study were to investigate the endogenous metabolic changes induced by AFB1 exposure and to further our understanding of AFB1 toxicity at the systemic level.
Materials and methods
Animals and sample collection
All animals involved in this study were cared for according to the principles of the Chinese Academy of Agricultural Sciences Animal Care and Use Committee. Twenty lactating Laoshan dairy goats (193 ± 14 d in milk, 1·36 ± 0·4 kg of milk/d, and 3·4 ± 1·0 parity) were grouped by parity (primiparous or multiparous), average daily milk yield, and day in milk and randomly assigned to one of two treatments (n = 10), according to a randomized block design. The treatments were 0 (control) and 50 µg/kg AFB1 (AFB1). The AFB1 was purchased from Sigma-Aldrich (St. Louis, MO, USA) and had a purity of 98%. The experimental period consisted of one week for adaptation to the environment followed by two weeks for the feeding of mycotoxins. The goats were assigned to a free stall barn with access to fresh water, and they were fed the total mixed ration (TMR). Pure AFB1 was separately dissolved in methanol. Pelleted concentrate was used as the mycotoxin carrier, and it was top-dressed in equal portions on the TMR at the 5 : 30 a.m. feeding.
Blood samples were collected from each goat before the morning feeding via jugular venipuncture on 14 d. Nine millilitres of blood were collected in vacuum tubes without anticoagulant, centrifuged at 3000 × g for 15 min at 4 °C to obtain serum, and stored at −80 °C for later analysis.
Sample preparation and 1H NMR spectroscopy
Serum samples were prepared by mixing 300 µl of serum with 100 µl of D2O and 200 µl of phosphate buffer saline solution (K2HPO4/NaH2PO4, pH 7·4). After vortexing and centrifugation at 13 000 rpm for 10 min, the supernatant was pipetted into 5 mm NMR tubes for NMR analysis.
A water-presaturated Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence (recycle delay-90°-(τ−180°-τ)n-acquisition, where τ = 400 µs and n = 400) was employed to retain NMR signals from small metabolites and to attenuate NMR signals from macromolecules with spin-spin relaxation delays (2nτ) of 320 ms. CPMG spectra with a 2 ms sample time and a 3 s relaxation delay between successive scans (64 scans per spectrum, 32 K data points, spectral width 8000 Hz) were recorded using a Varian INOVA-600 spectrometer with a pulsed field gradient and three resonance probes.
Metabolites were identified by inserting the experimental spectra into the Chenomx spectral database (Edmonton, AB, Canada) for comparison with the spectra of standard compounds.
NMR spectrum processing and multivariate data analysis
After manual corrections of both the phase and baseline, the chemical shifts were referenced internally to TSP (at δ = 0·0 ppm). The chemical shift region from 0·4 to 6·0 was segmented and integrated with equal widths of 0·01 ppm. All spectral segments were normalised to the total spectral area in Excel for multivariate data analysis.
Multivariate data analysis was carried out with SIMCAP+ software (version 13.0, Umetrics, Sweden). Prior to multivariate analysis, centre scaling was applied to 1H NMR data to reduce noise and artifacts in the models. Principal Component Analysis (PCA) was initially carried out on the NMR data to generate an overview. Partial least squares discriminant analysis (PLS-DA) models were applied to validate the model against overfitting through 999 random permutation tests. Discriminating variables were selected according to the loading plots, jack-knifed-based confidence intervals, variable importance in projection values (VIP > 1), and raw data plots in the orthogonal partial least squares discriminate analysis (OPLS-DA) models (Tian et al. Reference Tian, Wang, Zheng, Cheng, Li, Zhang and Wang2015). Furthermore, independent t-tests (P < 0·05) (SPSS version 13.0) were used to determine if the differences between the concentrations of candidate biomarkers obtained from OPLS-DA of both the control and AFB1 groups were statistically significant.
Results
Comparison of the metabolic profiles of the control and AFB1 groups
PCA models with good cluster of both the control and AFB1 groups and no strong outliers were observed (Fig. 1a). This presented apparent groupings, with only partial overlapping between the control and AFB1 data. The loading plots (Fig. 1b) reflected changes of some endogenous metabolites. To obtain improved model transparency and interpretability, OPLS-DA's for chemometric analysis were further performed to screen for differential variables. These showed a clear separation between the control and AFB1 groups without any overlap (Fig. 1c), which produced one predictive component and two orthogonal components with satisfactory modelling and predictive abilities R 2(X) = 41·9%, R 2(Y) = 91·3%, and Q 2(cum) = 76·9%. Loading plots in Fig. 1d showed changes of significantly altered endogenous metabolites. To avoid model overfitting, a default of seven rounds of cross-validation across three components was performed. Validation with 999 random permutation tests produced intercepts of R 2(Y) = 0·396 and Q 2 = −0·244 for the data (Supplementary Fig. S1). Overall, the results indicated that the predictive capabilities of the OPLS-DA models of the NMR data were reliable.
Fig. 1. (a) Score plot for the PCA model of the control and AFB1 groups. (b) Loading plots of centre-scaled 1H NMR data in unsupervised PCA model. (c) Score plot for the OPLS-DA model of control and AFB1 groups. (d) Loading plots of centre-scaled 1H NMR data in supervised OPLS-DA.
Identification of metabolite candidates
1H NMR analyses identified a total of 17 metabolic candidates (Table 1), which were classified into carbohydrate, amino acid, and lipid metabolites, suggesting that these metabolic pathways were altered in the AFB1 group. Supplementary Fig. S2 shows the signal 1H assignment of identified candidate metabolites.
Table 1. Candidate metabolite identified by † H NMR
† (d) doublet, (s) singlet, (dd) double doublet, (m) multiplet, (t) triplet
‡ P, independent t-test for control vs. AFB1
§ FC, fold change in the metabolite concentration (AFB1/control)
Metabolic alterations
Table 1 lists the metabolic alterations between the control and AFB1 groups. AFB1-induced alterations in carbohydrate-related metabolites were observed for lactate, glucose, citrate, and acetate. The concentrations of lactate and acetate were decreased by 0·71- to 0·89-fold in the AFB1 group compared with the control group (P < 0·01). The concentrations of glucose and citrate were increased by 1·21- to 2·89-fold in the AFB1 group compared with the control group (P < 0·01). Citrulline, leucine/isoleucine, valine, glycine, creatine, and betaine are metabolites related to amino acids; the former three were decreased by 0·84- to 0·91-fold in the AFB1 group compared with the control group (P < 0·05), and the latter three were increased by 1·32- to 1·64-fold (P < 0·01). The concentrations of acetone, acetoacetate, β-hydroxybutyrate, choline, phosphatidylcholine (PC)/ glycerophosphoryl choline (GPC), lipoprotein, and N-acetyl glycoproteins are lipid-related metabolites; the former two were increased by 1·27- to 1·51-fold in the AFB1 group compared with the control group (P < 0·01), and the latter two were decreased by 0·77- to 0·84-fold in the AFB1 group compared with the control group (P < 0·05).
Discussion
AFB1's influence on liver damage
Choline, PC, and GPC are essential elements for structural integrity of cell membranes. The significant decreases in serum levels of choline, PC, and GPC in AFB1 treated goats indicated that AFB1 causes liver damage, which was in line with the decrease of lipoprotein and N-acetyl glycoproteins in the serum. As an organic osmolyte, betaine can be excreted into serum to maintain osmolarity and protect against cell damage in hypotonic conditions when hepatocytes swell and lead to the accumulation of intracytoplasmic fluids in cells (Zerbst-Boroffka et al. Reference Zerbst-Boroffka, Kamaltynow, Harjes, Kinne-Saffran and Gross2005). The observation of elevated levels of betaine in the current study suggested that both the swelling of hepatocytes and subsequent liver damage had occurred.
Previous evidence from gene expression also supports these results. Several genes that were encoding cytoskeletal organisation were up-regulated in both middle- (0·75 mg/kg body weight) and high-dose (1·5 mg/kg body weight) AFB1 groups of rats, including Rhob in both groups and Anxa1, Anxa2, Anxa4, Anxa5, Anxa7, and Arpc1b in the high-dose group (Lu et al. Reference Lu, Hu, Shao, Tian, Jin, Jin, Ji and Fan2013). Additional supportive evidence of liver damage induced by AFB1 was found in rats that were dosed with 0·24 mg/(kg/day) of AFB1 for 14 d, where genes encoding the cytoskeletal constituent were up-regulated, including CRYAB ɑ B-crystallin, RhoB, SPTA2 spectrin ɑ chain (fodrin ɑ chain), and TMSB 10 thymosin β−10 (Ellinger-Ziegelbauer et al. Reference Ellinger-Ziegelbauer, Stuart, Wahle, Bomann and Ahr2004).
AFB1's impact on carbohydrate metabolism
An increase in citrate excretion together with decreased lactate excretion was observed with 1H NMR detection, which suggested impairments in both Cori and Krebs cycle metabolism in periods of altered energy production, which may be linked to damaged mitochondrial function. The major function of electron transport is ATP synthesis in mitochondrion. The depletion of ATP that was seen resulted from impaired energy metabolism. A previous investigation also reported that ATP synthesis pathways are more affected in chicks fed 2 mg of AFB1/kg, leading to decreased energy production and the genes being down-regulated, such as in the adrenodoxin mitochondrial precursor, cytochrome P450 2C5, cytochrome P450 2P3, and nicotinamide adenine dinucleotide phosphate-cytochrome b5 reductase (Yarru et al. Reference Yarru, Settivari, Antoniou, Ledoux and Rottinghaus2009). Consistent with the findings in broilers (Tessari et al. Reference Tessari, Kobashigawa, Cardoso, Ledoux, Rottinghaus and Oliveira2010) and ducks (He et al. Reference He, Zhang, Chen, Ding, Feng and Ao2013), our goats that were fed diet supplements of 50 µg/kg of AFB1 had a 1·21-fold changes of the glucose level in serum, indicating that liver function had been damaged and that a consequence of AFB1 exposure is an accelerated rate of glycogenolysis. Such depletion of hepatic glucose has also been observed in rats dosed with 0·32 mg/(kg body weight/day) of AFB1 (Zhang et al. Reference Zhang, Ye, An, Tian, Wang and Tang2011).
AFB1's impacts on lipid metabolisms
A previous investigation found increases in the relative liver weights of chicks that were fed aflatoxin (Ortatatli & Oğuz, Reference Ortatatli and Oğuz2001). In this study, we observed decreased levels of β-hydroxybutyrate and acetate in the serum after AFB1 treatments, indicating that AFB1 exposure alters the lipid metabolism of goats. These results were consistent with the findings of Yarru et al. (Reference Yarru, Settivari, Antoniou, Ledoux and Rottinghaus2009), who found that genes involved in fatty acid metabolism were downregulated in chicks that were fed 2 mg/kg of AFB1. Acetoacetate and acetone are the products of the β-oxidation of fatty acid in mitochondria, hence, their elevations in this study suggested the promotion of β-oxidation of fatty acid with AFB1 exposure. In agreement with our findings, a previous investigation found that AFB1 exposure caused up-regulation of peroxisome proliferator-activated receptor R, a key regulator of lipid metabolism (Yarru et al. Reference Yarru, Settivari, Antoniou, Ledoux and Rottinghaus2009).
AFB1's impact on amino acid metabolism
In this study, AFB1 induced an increase in the levels of citrulline, leucine/isoleucine, and valine and a decrease in the level of glycine, which implies that AFB1 altered hepatic amino acid metabolism. Moreover, a noticeable increase in creatine levels was observed in the serum of AFB1-treated goats. Phosphocreatine, an energy reservoir, can be rapidly converted to creatine by creatine kinase and the release of adenosine triphosphate (ATP) and creatine in response to high energy demands (Diao et al. Reference Diao, Zhao, Guan, Zheng, Chen, Yang, Lin, Chen and Gao2014). Suppression of the tricarboxylic acid (TCA) cycle caused by AFB1 exposure will inevitably result in insufficient energy production. Hence, the accumulated creatine in the serum of AFB1 exposed goats may be the result of the accelerated conversion of phosphocreatine to creatine to meet this energy demand. The decreased serum levels of citrulline implied that the urea cycle was interrupted by AFB1 because it is the intermediate metabolite in the urea cycle. This is inconsistent with the AFB1-induced down-regulation of gene expression of carbamoyl-phosphate synthetase I in mice (Zhuang et al. Reference Zhuang, Lin, Yang, Yuan and Wang2014).
Conclusion
In conclusion, dairy goats’ exposure to low doses of AFB1 caused significant alterations in several metabolic pathways, including lipid oxidation, TCA cycle, and carbohydrate and amino acid metabolism. Our results also show that 1H NMR-based metabolomic analysis of the serum provided an effective method for understanding AFBI's adverse effects on goats’ health at low levels as well as a method for preventing AFM1 contamination of milk.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0022029917000590.
This work was supported financially by Agro-scientific Research in the Public Interest (201403071), Introduction of International Advanced Agricultural Science and Technology Program (2015-Z12; 2016-X37) and The Agricultural Science and Technology Innovation Program (ASTIP-IAS12).
