Mycotoxins are toxic secondary metabolites of different molds, such as Aspergillus
spp. or Alternaria
spp., and often found in animal feeds. They impair animal health by manifold modes of action, causing hepatotoxic, nephrotoxic, immunomodulatory, genotoxic, and neurotoxic effects as well as reproductive and developmental disorders [1
]. During the last decade, the intestine has moved into the spotlight of mycotoxin research. It represents the first barrier to these feed contaminants and is often exposed to higher mycotoxin concentrations than other body tissues. Here, mycotoxins do not only affect digestion and nutrient uptake, but also intestinal histomorphology, gut barrier integrity, mucin production, microbiota composition, and the local immune system [2
Due to their frequent occurrence and negative impact on animal health, many countries have established regulations for mycotoxins in feed. In the European Union (EU), maximum limits are in place for aflatoxin B1 (AFB1) and ergot alkaloids [4
], while guidance levels have been set for deoxynivalenol (DON), zearalenone (ZEN), ochratoxin A (OTA) and the sum of fumonisin B1 (FB1) and fumonsin B2 (FB2) [5
]. These regulations neither take the presence of multiple mycotoxins into account, nor the occurrence of so-called emerging mycotoxins. This heterogenous group of fungal metabolites is not clearly defined, but commonly referred to as “mycotoxins, which are neither routinely determined, nor legislatively regulated; however, the evidence of their incidence is rapidly increasing” [6
]. Proper risk assessment of emerging mycotoxins, e.g., enniatins, culmorins, beauvericin, moniliformin, roquefortine C or fusaric acid, is challenging, as data on toxicity and occurrence are still scarce [7
Forages are especially prone to contamination by emerging mycotoxins [8
]. Fresh, dried and ensiled forages are important components of ruminant diets, representing 50–75% of the total diet [10
]. Ensiling describes the preservation of green forage by lactic fermentation under anaerobic conditions and shows geographic variations concerning the quantity and type of silage produced [11
]. In the EU-28 alone, approximately 2.4 million tons of green maize, which is mainly grown for silage, were harvested in 2018 [12
]. Silages can contain a wide range of mycotoxins, that originate either from pre-harvest contamination, or from spoilage with (acid-tolerant and micro-aerobe) toxigenic fungi during storage [8
]. Hence, ruminants might be exposed to a plethora of mycotoxins, in particular compared to chicken or swine, which have less diverse diets [9
]. However, this risk has been poorly addressed so far, and the need for thorough mycotoxin monitoring in ruminant forages has been highlighted only recently [9
In general, ruminants are considered to be less susceptible to mycotoxins than other livestock species, mainly because their ruminal microflora is capable of degrading certain mycotoxins to less toxic metabolites [8
]. Most prominently, DON is extensively metabolized to de-epoxy-deoxynivalenol (DOM-1), reaching conversion rates of up to 81–99% in dairy cattle [13
]. The close connection between a functional ruminal microflora and DON toxicity was impressively depicted by Valgaeren et al. [15
]. Driven by clinical cases of DON toxicosis in 2- to 3-month-old beef calves, the authors showed that the oral bioavailability of DON is markedly increased in non-ruminating calves (50.7%) compared to ruminating calves (4.1%). Moreover, it was recently demonstrated that a low ruminal pH-value can impair the degradation of DON, NIV, ZEN, and enniatin B (ENNB) in vitro [16
]. Especially in the light of subacute rumen acidosis, one of the most important nutritional diseases in dairy cattle [17
], these findings are of significant practical relevance. In addition, certain mycotoxins, e.g., ENNB [18
], exert antimicrobial activity. It has therefore been suggested that such mycotoxins might alter the ruminal microflora and its degradation capacity [8
]. Finally, some mycotoxins hardly undergo ruminal metabolism [8
]. For example, limited degradation of 10–18% was reported for FB1 [19
]. Hence, major amounts of mycotoxins might reach the small intestine and affect gut health both in non-ruminating calves and dairy cattle.
The aims of our study were twofold. First, we investigated the mycotoxin exposure of dairy cattle. To this end, a total of 158 maize silage samples were collected in Europe and analyzed for 61 mycotoxins, including regulated as well as emerging mycotoxins. Second, an in vitro model using calf small intestinal epithelial cells B (CIEB) was established to assess the cytotoxicity of DON, NIV, FB1 and ENNB. Thus, our study does not only deliver comprehensive mycotoxin occurrence data, but also new toxicological information regarding the relevance of mycotoxins for bovine gut health, a previously neglected target in ruminants.
Mycotoxin occurrence is influenced by multiple factors, including plant species and variety, region, temperature, humidity, insect damage, storage conditions, and other agricultural practices [22
]. Our survey focused on the presence of mycotoxins in maize silage, because this feed component can be the main source for dietary mycotoxin intake in dairy cattle [23
]. Since sample numbers per country and/or year were limited in our survey, definite conclusions on regional or yearly trends of mycotoxin occurrence were not justified and therefore omitted. Respective information can be retrieved from other excellent feed surveys [24
] and reviews [9
]. Similar to the approach of Storm et al. [25
], mycotoxin concentrations were expressed as µg/kg fresh weight except for the comparison with EU maximum/levels, for which levels were normalized to a dry matter content of 88%. Because literature reports do not uniformly express mycotoxin concentrations in silage (using either fresh or dried weight), the suffix “fresh weight” is used in the following whenever clearly indicated in the respective study, or when samples were not dried prior to analysis.
In 98.8% of silage samples at least one mycotoxin was detected. The top five positions in terms of prevalence were all held by emerging mycotoxins, namely EMO, CUL, ENNB1, ENNB, and BEA. Although data on the presence of emerging mycotoxins in feed are scarce, high incidences of enniatins and BEA have been described previously. For example, ENNB1, ENNB, and BEA were found in 97%, 90%, and 100% of maize silages collected in Poland, respectively [26
]. Reported median values (6.0–20.9 µg/kg fresh weight) were in a similar range in our study. In silage samples from Spain [27
], ENNB showed yearly variations in prevalence (31–72%) with higher average concentrations (151–163 µg/kg). In contrast, moderate incidences of around 25% for ENNB [25
] and BEA [28
] were reported in Denmark. Differences between studies might stem from distinct fungal contamination patterns and/or variations in methodology (e.g., sampling procedure, limits of detection). Reports on the toxicity of these Fusarium
toxins in ruminants are completely lacking so far [29
]. In this respect, the described antimicrobial activity of enniatins and BEA, potentially affecting the composition and function of the rumen microbiota, might be of special interest. In addition, ENNB and BEA were demonstrated to impair the barrier function in intestinal porcine enterocytes (IPEC-J2; [30
]). Since a certain proportion of ENNB might by-pass the rumen [16
], negative effects on the bovine gut cannot be excluded.
To the best of our knowledge, the occurrence of EMO and CUL in European maize silages has not been addressed yet. In line with our data, a survey conducted in Israel showed high prevalence of EMO in maize silages (100%; [31
]). For CUL, results deviate from our study, mostly in terms of incidence (6.6% versus 79.1%) but also concerning median values obtained (46 µg/kg fresh weight versus 190 µg/kg). This mycotoxin has recently caught scientific attention because of its potency to inhibit DON glucuronidation [32
], and we confirmed the commonly observed co-occurrence of CUL and DON [7
] for maize silages. Still, the relevance of CUL for dairy cattle remains debatable, as metabolization to DOM-1 is the primary detoxification pathway for DON in ruminants. Another emerging mycotoxin that has gained certain interest is fusaric acid. Shimshoni et al. [31
], authors of the aforementioned survey in Israel, pointed out both its high prevalence and concentration in maize silage. Evaluated on a larger sample size and in a different region, our findings corroborate a certain relevance for fusaric acid (detected in nearly one quarter of silages, maximum concentration of 4,120 µg/kg fresh weight). Concerns for bovine health were related to the growth inhibition of important rumen microorganisms and the toxin’s potential carry over to milk [31
]. However, like for other emerging mycotoxins, toxicodynamic and toxicokinetic studies are warranted to verify these assumptions and to elucidate the role of fusaric acid for food safety.
FB1+FB2 were detected in approximately one third of the samples, albeit at low concentrations. As unveiled by Latorre et al. [33
], the majority of fumonisins in maize silage are present in a modified form. These so-called “hidden fumonisins” escape routine analysis, but are expected to be released upon mammalian digestion [34
]. For assessing the total fumonisin burden, alkaline hydrolysis of samples is required [33
]. This was not performed in the present case and thus represents a limitation of our study. Similarly, total exposure to type-A or -B trichothecenes is underestimated in surveys that do not account for acetylated or modified forms. In our study, DON showed a high prevalence of 67.7% with moderate median concentrations of 303 µg/kg fresh weight. In the past, higher average DON values of 1,629 µg/kg fresh weight [25
] or 854–1316 µg/kg [10
] were monitored, and incidences varied substantially from 6.1–86% [25
]. While average molar DON-3-Glc/DON percentages of 20% were proposed for cereals [36
], we found markedly lower values of 2.72%. Further studies are necessary to assess whether this observation is related to the commodity maize silage as such or merely to our sample set. The same applies to our findings on NIV-3-Glc and HT2–3-Glc, both showing negligible prevalence. Although this indicates that NIV-3-Glc does not contribute significantly to the total NIV burden of dairy cattle, the prominent prevalence of the parent toxin (59.5%) must be underlined. Maximum NIV values exceeded previously reported data [25
], revealing that the NIV exposure can be extremely high for individual dairy cattle herds.
ZEN was the only mycotoxin found at levels above the EU maximum/guidance limits [4
], with 5.1% of samples exceeding ≥ 2,000 µg/kg ZEN. In most mycotoxin surveys, maize silages complied to the EU regulations [10
], whereas Dangac et al. [27
] reported 1.4% of samples exceeding the recommended maximum levels for ZEN. It should be noted that a different limit was employed in that study (500 µg/kg ZEN for complete feedstuffs), which hampers a direct comparison of results. Still, data emphasize the need to monitor ZEN in maize silage and to control its formation pre- and postharvest. This is especially important in the light of potential synergistic effects with other mycotoxins. Naturally, ZEN often co-occurred with DON (63.3%). While exposure to diets co-contaminated with ZEN and DON did not affect the performance of dairy cows [38
], alterations of health-related blood parameters were observed by Dänicke et al. [39
]. In addition, authors suggested an influence on ketogenesis at the cellular level. Clearly, more studies are needed to decipher the interactions of DON and ZEN in ruminants. The same is valid for other mycotoxin combinations.
Co-occurrence of mycotoxins might be of relevance for animal health even at comparably low concentrations. As summarized by Chehli et al. [40
], the type and intensity of mycotoxin interactions can vary dose-dependently. Our study confirmed that mycotoxin co-occurrence in feed is rather the rule than the exception. Strikingly, silage samples contained 13 mycotoxins on average, and in 87% of samples more than five mycotoxins were found. These high values are partly attributed to the broad palette of mycotoxins tested in our study, and therefore, further expand existing knowledge on mycotoxin co-contamination in maize silage (e.g., Refs. [9
Next, we investigated the impact of silage mycotoxins on bovine gut health. Based on our survey results, we focused on Fusarium
toxins and assessed the cytotoxic potential of DON, NIV, FB1, and ENNB on bovine intestinal cells. Toxins were selected due to their high prevalence (DON, ENNB), maximum concentrations (NIV) or ruminal stability even under physiological conditions (FB1; [20
]). The toxicity of mycotoxins in bovine intestinal cells is currently unknown, mainly because of two reasons. First, the rumen microbiota was long thought to neutralize the toxicity of mycotoxins. However, recent studies indicate that the ruminal degradation capacity might be impaired under specific conditions, such as altered ruminal development in calves [15
] or rumen acidosis [16
]. Second, the small number of commercially available bovine lines restrains research in this field.
Since CIEB are not widely used, we first confirmed the species identity and the absence of mycoplasma contamination. Although these aspects are of paramount importance for reliable and reproducible in vitro results, they are often neglected. Mycoplasma contamination can alter the properties of cell lines, and infected CIEB were described to exhibit low viability and poor growth [41
]. In the same study, authors reported misidentification of three out of eight tested cell lines. The dimension of this issue is even more striking when retrieving information from the International Cell Line Authentication Committee, which has documented 451 false identified cell lines [42
]. Resources wasted in the last 50 years due to misidentification of cell lines, stemming from cross-contamination, wrongly labelled samples or inadequate protocols, can only be estimated [43
]. Consequently, increased attention should be paid to adequate quality controls for in vitro experiments, also in mycotoxin research.
Further characterization of CIEB was performed by immunofluorescence staining. CIEB showed a positive reaction for cytokeratins and villin. Cytokeratin proteins, which are characteristic components of the cytoskeleton, are commonly used for identification of epithelial cells [44
]. Villin is an actin-binding protein in the microvilli of epithelial cell [45
]. Expression of both proteins has been used to verify the intestinal epithelial nature of bovine cells before [46
]. Besides, CIEB were immuno-positive for vimentin. This protein is a typical marker for non-epithelial cells, such as fibroblasts [47
]. However, unequivocal identification of fibroblasts remains challenging. For example, the expression of vimentin was reported for the intestinal porcine epithelial cell line (IPEC-J2; [48
]). Another study even excluded the presence of fibroblasts in cells isolated from calf intestine although they showed a positive reaction for vimentin [49
]. It seems that the expression of vimentin is not a unique property of mesenchymal cells but can also be found in intestinal epithelial cells and should be evaluated in combination with the presence/absence of cytokeratin expression. Altogether, our immunohistochemistry results asserted the epithelial intestinal origin of CIEBs.
Absolute IC50 values of mycotoxins were calculated based on viability tests performed with three different assays (WST-1, SRB, NR). Independent of the assay, NIV was the most cytotoxic mycotoxin (IC50 0.8–1.0 µM), closely followed by DON (1.2–3.6 µM). The higher comparative cytotoxicity of NIV is in accordance to experiments performed in human (epithelial colorectal adenocarcinoma cells, Caco-2; [50
]) and porcine intestinal cells (IPEC-1, IPEC-J2; [51
]). Likewise, the absolute IC50 values obtained for NIV and DON in CIEB are in a similar range as reported previously. For example, IC50 values for NIV were 0.9–2.1 µM in Caco-2, IPEC-1 and IPEC-J2 cells [50
], and 0.9–3.6 for DON [50
]. Opposed to that, individual studies found higher IC50 values, e.g., 6.9 µM for NIV [54
] or up to 44.8 µM for DON [55
]. Differences between studies can derive from experimental conditions, such as cultivation medium, tested concentration range, exposure period, chosen endpoint, calculation of IC50 values, or differentiation status of cells [40
]. Overall, data indicate that CIEB are at least as sensitive to NIV and DON as human or porcine intestinal cells.
For ENNB, a higher cytotoxicity compared to NIV [54
] and DON [56
] was observed in Caco-2, which could not be confirmed for CIEB. Interestingly, the most sensitive IC50 value for ENNB (4.0 µM) was generated by the NR assay, which measures lysosomal activity. Indeed, destabilization of lysosomes has been suggested as an upstream event of ENNB-induced cell death [57
]. As IC50 values after incubation periods of up to 48 h varied strongly in Caco-2 (2.1 to > 30 µM; [18
]), comparison of results is challenging. Still, in line with the present study, it was reported that the NR assay yields lower IC50 values for ENNB than assays measuring metabolic activity [57
]. Since mitochondria are one of the major cellular targets of enniatins [18
], we originally assumed a strong response in the WST-1 assay. However, among other effects on these cell organelles, enniatins induced swelling of rat liver mitochondria [59
]. Interestingly, the same phenomenon was described for IPEC-J2 cells exposed to DON, and here the comparably weaker cytotoxic response assessed by the WST-1 assay was partly attributed to alterations of the mitochondrial morphology and metabolic activity [30
]. Although further mechanistic studies are needed, cumulative data suggest that the metabolic activity does not represent the most sensitive endpoint for cytotoxicity assessment of ENNB. For the SRB assay, calculation of an IC50 value was not possible. To the best of our knowledge, no other study has employed this test to determine the cytotoxicity of ENNB so far. Hence, we cannot conclude whether the total protein content is the least sensitive endpoint for this mycotoxin or whether this finding is rather limited to our experimental conditions. Concordant with Springler et al. [55
], our study underlines the importance of multi-parameter analysis in the cytotoxicity assessment of mycotoxins.
FB1 showed the lowest cytotoxicity among the toxins tested in our study. Considering previous studies demonstrating minor cytotoxicity of this mycotoxin in Caco-2 [60
] and IPEC-J2 [52
], these results are not surprising. Still, it should be noted that IC50 values obtained in CIEB are markedly lower than the ones reported previously for other intestinal cells (if computable at all). This might be partly related to the narrow concentration range tested in some of the studies [52
]. FB1-induced effects on the intestine were proposed to originate from disruption of the sphingolipid metabolism which causes intracellular accumulation of the sphingoid bases sphinganine (Sa) and sphingosine (So) [65
]. In line with reports addressing intestinal tissues/cells from monogastric livestock species [66
], a dose-dependent increase of the Sa/So ratio was observed in CIEB. Compared to Loiseau et al. [66
], who found a significant elevation of the Sa/So ratio after 48 h of exposure to 100 µM FB1 in IPEC-J2, CIEB reacted to lower toxin concentrations, reaching statistical significance at 6.25 µM FB1. Yet, absolute Sa/So values were smaller in our study, which might be explained by the type of matrix used for analysis (cell extract [66
] or supernatant).
Although ZEN showed high prevalence in silage samples and was the only mycotoxin exceeding the EU guidance levels, we did not include this compound in our cytotoxicity experiments. This decision was mainly based on the primary mode of action of ZEN, which is the activation of estrogen receptors [68
]. Compared to other mycotoxins, the effects of ZEN on the intestine are less detrimental [69
]. For example, IC50 values for ZEN obtained by measuring metabolic activity in Caco-2 were 313 µM [70
] and 25 µM [62
] after 48 and 72 h of incubation, respectively, and thus even higher than those observed for FB1 in the same experiments. However, increased sensitivity of CIEB to ZEN cannot be ruled out at this stage and should be addressed in future studies. In vitro models represent an essential tool to unravel the toxicological relevance and mode of action of substances. Yet, direct extrapolation to in vivo conditions is often limited, mainly because in-vitro experiments cannot fully reflect the complexity of an intact organism [40
]. In an attempt to compare concentrations used in our in vitro experiment to mycotoxin levels in dairy feed, we used the dataset provided by Seeling et al. [14
]. In this study, 14 duodenal fistulated cows were exposed to DON-contaminated feed, which allowed the assessment of the toxin’s duodenal flow. On average, 1.3% on ingested DON reached the duodenum in unmetabolized form. This low proportion partly stemmed from ruminal absorption of the toxin, but mostly from metabolization to DOM-1 (94–99%). Calculating with this percentage, the IC50 value for DON in CIEB (356 µg/L; WST-1) theoretically corresponds to a feed concentration of approximately 27,400 µg/kg. Although this value exceeds the maximum DON levels detected in fresh maize silage by a factor of ten, it should not be overlooked that minor changes in the ruminal degradation capacity would have a marked impact on the outcome of this estimation. As such, it highlights the practical relevance of our findings.