Thiouracil-Forming Bacteria Identified and Characterized upon Porcine In Vitro Digestion of Brassicaceae Feed

Abstract

In recent years, the frequent detection of the banned thyreostat thiouracil (TU) in livestock urine has been related to endogenous TU formation following digestion of glucosinolate-rich Brassicaceae crops. Recently, it was demonstrated that, upon in vitro digestion of Brassicaceae, fecal bacteria induce TU detection in livestock (porcine livestock > bovines). Therefore, the present study was intended to isolate and identify bacteria involved in this intestinal TU formation upon Brassicaceae digestion and to gain more insight into the underlying mechanism in porcine livestock. Twenty porcine fecal inocula (gilts and multiparous sows) were assessed through static in vitro colonic-digestion simulations with rapeseed. After derivatization and extraction of the fecal suspensions, TU was analyzed using liquid chromatography-tandem mass spectrometry (LC-MS²). On average, lower TU concentrations were observed in fecal colonic simulations in gilts (8.35 ng g⁻¹ rapeseed ± 3.42 [mean ± standard deviation]) than in multiparous sows (52.63 ng g⁻¹ ± 16.17), which correlates with maturation of the gut microbial population with age. Further exploration of the mechanism showed cell-dependent activity of the microbial conversion and sustained TU-forming activity after subjection of the fecal inoculum to moderate heat over a time span of up to 30 min. Finally, nine TU-producing bacterial species were successfully isolated and identified by a combination of biochemical and molecular techniques as Escherichia coli (n = 5), Lactobacillus reuteri (n = 2), Enterococcus faecium (n = 1), and Salmonella enterica subsp. arizonae (n = 1). This report demonstrates that endogenous formation of TU is Brassicaceae induced and occurs under colonic conditions most likely through myrosinase-like enzyme activity expressed by different common intestinal bacterial species.

INTRODUCTION

The continuous challenge of maintaining a safe and healthy food chain has led to some remarkable findings regarding the occurrence of controlled or forbidden residues in animal husbandry. Generally, it was believed that the detection of banned substances in livestock could be attributed to fraudulent administration. But, in the last decade, this belief has been called into question by recurring discrepancies showing that certain forbidden substances, such as the thyreostat thiouracil (TU), may occur in animal matrices without a history of illicit administration (1, 2). This facilitates the incidence of possible false accusations of illicit use due to noncompliant TU results during national control plan campaigns, because, in this case in particular, no method of differentiation between exogenous administration and endogenous formation is yet routinely applicable. The use of TU, a thyreostatic drug of synthetic origin, in animal production was forbidden by the European Union more than 3 decades ago (3). The equivocal reasons behind this decision were twofold. On the one hand, fraudulent slaughtering-weight gain could be achieved through TU administration, which causes water retention in edible tissues and gut and inevitably leads to a deterioration in meat quality. On the other hand, the potential hazard for public health safety was taken into account, since carcinogenic (4) and teratogenic (5, 6, 7) effects had been described. Consequently, the European Union demands that its member states perform thyreostat analysis of diverse matrices (e.g., urine and meat). As a guideline, a European Minimum Required Performance Limit (MRPL) was stipulated for the analytical detection of thyreostats, ensuring detection of 100 μg liter⁻¹ or kg⁻¹ in all matrices (8). In recent years, analytical detection methods have been capable of reaching detection limits below 10 μg liters⁻¹ (9, 10, 11). As a result, low-level (<10 μg liter⁻¹) TU concentrations have frequently been detected in national control plans in urine of untreated livestock (2, 12, 13). Because of this, the European Union Reference Laboratory specified a recommended concentration (RC) of 10 μg liter⁻¹ or kg⁻¹ (urine and thyroid) for thiouracil (14), which may be considered a technical guideline for laboratories involved in European Union control plans but has no legal constraining power.

Pinel et al. (1) were the first to describe the possibility of a feed-related origin of TU in livestock. More specifically, they discovered that digestion of Brassicaceae crops resulted in detection of TU (<10 μg liter⁻¹) in bovine urine. Brassicaceae crops (e.g., rapeseed and cabbage) have been widely used in livestock feedstuffs for their cheap and high-quality nitrogen content, especially during winter. Another well-known characteristic is, however, their high glucosinolate content (15). Upon plant disruption (e.g., by chewing), glucosinolates, located in the idioblast cells of the plant, are freed from their vacuoles together with the plant-derived enzyme myrosinase (ß-thioglucosidase; EC 3.2.1.147), which is released from the parenchymatous tissue (15), causing the formation of various breakdown products (oxazolidine-thiones, nitriles, epithionitriles, thiocyanates, isothiocyanates, and thiourea) depending on pH (16, 17). Some of these breakdown products are natural goitrogenic substances affecting the thyroid function, e.g., oxazolidine-thiones (goitrin), thiocyanates, and thiourea (18). Negative effects of glucosinolates on animal production are proportional to their dietary concentration (19), and the exploitability of glucosinolate-rich crops in the feed industry is consequently limited (15). Subsequently, this led to a minimization of glucosinolate content of Brassicaceae crops with the development of “00”-type plants, which have low glucosinolate content (maximum of 25 μmol g⁻¹ at a moisture content level of 9%) and erucic acid content (<2% of the total fatty acid content) (20).

In order to confirm the possible link between glucosinolates and TU formation, research by Vanden Bussche et al. (21) demonstrated that the interference of plant-derived myrosinase was a necessity for the (in vitro) hydrolysis of glucosinolate-rich Brassicaceae food and feed matrices into TU. Additionally, the natural occurrence of TU in the urine of various domesticated animals and in that of humans was investigated. The animals displayed traces of TU below 10 μg liter⁻¹ without any diet control (2); as for humans, for whom the influence of a controlled Brassicaceae diet was investigated, low-level TU was retrieved in 66.7% of the samples.

Besides plant-derived myrosinase hydrolysis, bacterial hydrolysis resulting from intestinal microbiota has also been linked to glucosinolate degradation during digestion (22, 23, 24). For example, the hydrolysis of sinigrin, an aliphatic glucosinolate, can be mediated by the human intestinal bacterium Bacteroides thetaiotaomicron in gnotobiotic mice (25). Several bacterial strains in animals have also been investigated for their glucosinolate-degrading activity. For example, a Lactobacillus strain was capable of inducing myrosinase-like enzymatic activity in vivo (26) and the Lactobacillus agilis R16 strain could degrade sinigrin and glucosinolates in brown mustard seed extracts (27). A study by Kiebooms et al. (28) was the first to demonstrate in vitro that TU formation in bovine and porcine livestock was influenced by a bacterial component and identified the large intestine, with its high numbers and diversity of microbes (29, 30), as the main transformation site. Concentrations up to 80 μg kg⁻¹ TU were registered in in vitro livestock digestion simulations with Brassicaceae.

Consequently, the focus of the present study was to further investigate the potential and identity of the intestinal microbiota as well as the mechanisms involved in TU formation upon Brassicaceae digestion in livestock by conducting static in vitro colonic-digestion simulations in porcine livestock, because of their high TU-producing capacity (28).

MATERIALS AND METHODS

Reagents and chemicals.

The chemical standard 2-thiouracil (TU) was obtained from Sigma-Aldrich (St. Louis, MO). The deuterated internal standard for TU, 6-propyl-D₅-2-thiouracil (PTU-D₅), was obtained from Toronto Research Chemicals Inc. (North York, Ontario, Canada). Stock solutions of the chemical standards were prepared in methanol at a concentration of 200 ng μl⁻¹. Working solutions were diluted in methanol to 1 ng μl⁻¹ for PTU-D₅ and 1 ng μl⁻¹, 0.1 ng μl⁻¹, and 0.01 ng μl⁻¹ for TU. Solutions were stored in dark-glass receptacles at 4°C. Reagents were of analytical grade (VWR International/Merck, Darmstadt, Germany) when used for extraction and purification purposes and of Optima liquid chromatography-mass spectrometry (LC-MS) grade for LC-MS applications (Fisher Scientific United Kingdom, Loughborough, United Kingdom). The derivatization reagent, 3-iodobenzyl bromide (3-IBBr; Sigma-Aldrich, St. Louis, MO), was prepared extemporaneously (2 mg ml⁻¹ methanol). A phosphate buffer, consisting of 0.2 M Na₂HPO₄ and 0.2 M KH₂PO₄ (Merck, Darmstadt, Germany) in deionized water, was prepared and adjusted to pH 8 with 6 M HCl.

Feed and feces sampling.

Traditional rapeseed (Brassica napus L. partim Napoleon; Institute for Agriculture and Fisheries Research [ILVO], Melle, Belgium) was used for the in vitro digestion experiments as a substrate for high TU production upon bacterial fermentation. The large intestine in vitro digestion simulations required sampling of several porcine fecal inocula (FI). For this purpose, fresh fecal matter was collected upon defecation from 10 female gilts (<8 months old) waiting for their first insemination (31) and 10 adult multiparous female porcine livestock from the Institute of Agricultural and Fisheries Research (ILVO, Melle, Belgium) and Ghent University, Faculty of Veterinary Medicine (Ghent, Belgium). The animals were housed according to animal welfare requirements, keeping them on a standard antibiotic-free maintenance diet sustaining their general metabolism. The feces aliquots were quickly transported at room temperature to the laboratory where further processing was performed.

In vitro large intestine digestion. (i) Preparation of buffer, broth, and agar.

Digestion broth media, agar media, and phosphate-buffered saline were prepared in ultrapure water (UPW), processed in conformance with the manufacturing guidelines (Sartorius stedim Biotech GmbH, Göttingen, Germany), and autoclaved for 15 min at 121°C and 1 atm to ensure the absence of bacterial contamination and subsequent growth. All procedures following the autoclave procedure were performed in a laminar flow cabinet to prevent bacterial contamination. Autoclaved digestion broths, buffers, and agar plates were all stored tightly sealed at 4°C for no longer than 1 month.

Phosphate-buffered saline, wherein the fresh feces was suspended, contained K₂HPO₄ (8.8 g liter⁻¹), KH₂PO₄ (6.8 g liter⁻¹) (Merck, Darmstadt, Germany), and sodium thioglycolate (1 g liter⁻¹) (Sigma-Aldrich, Steinheim, Germany).

Various liquid growth media were used throughout the different trials depending on the organisms cultivated (e.g., whole flora, specific bacterial isolates, etc.). For general preculturing and cultivation of porcine microorganisms, brain heart infusion broth (BHI) (CM1135) (Oxoid Ltd., Hampshire, England) was used (28). Specialized media were used for the cultivation of specific microbial strains: Sabouraud dextrose liquid medium (SDx) (CM0147) for fungi and yeasts, Wilkins-Chalgren anaerobe broth (WCh) (CM0643) for anaerobes, MacConkey broth (MacC) (CM0005) for coliforms, and de Man Rogosa Sharpe broth (MRS) (CM0359) for lactic acid bacteria (Oxoid). Purified isolates were preserved as stocks containing tryptic soy broth (TSB) (CM131; Oxoid) and glycerol (1:3) (Analar Normapur VWR, Fontenay-sous-Bois, France) (bidistilled [99.5%]) at −80°C. In addition, the microorganisms were stored at −80°C in strain-specific media in cases of activity loss in general medium (e.g., MRS broth for lactic acid bacteria).

To improve anaerobic growth, l-cysteine (SAFC Supply Solutions, St. Louis, MO) (0.5 g liter⁻¹) was added (32) to liquid media, while broths were prereduced by boiling, which further reduced the oxygen content.

For counting or isolation purposes, agar media were made by adding bacteriological agar no. 1 (LP0011; Oxoid) to the liquid broth medium of choice, which was heated until boiling and subsequently autoclaved.

To count the bacteria present in fecal digestions, an automated spiral plater was used (EddyJet; IUL Instruments, Led Techno NV, Heusden-Zolder, Belgium) (100 μl per plate). Plates were incubated in sealed jars containing AnaeroGen (AN00035) sachets (Oxoid) (3.5 liters) in a quantity according to the size of the jar to attain anaerobicity.

For the liquid digestion media, an N₂-flushing system equipped with sterile filters and needles was applied to obtain anaerobicity in the sterile digestion flasks. In order to control its efficacy and to determine the required flushing time, dummy digestions were supplemented with resazurin (Sigma-Aldrich) (2 mg liter⁻¹) and subsequently flushed, causing extensive reduction of resazurin (blue) into resorufin (pink) and then into hydroresorufin (colorless) over time, once anaerobic conditions were achieved (33).

(ii) Digestion protocol.

The protocol applied for the static in vitro colonic-digestion simulations in the present study was adapted from the scientific literature on simulations of the gastrointestinal tract of humans and porcine livestock (28, 34, 35).

(a) Fecal and inoculum preculturing.

Fecal slurry was attained by addition of phosphate-buffered saline to the fresh feces (1/5 solid/liquid ratio) followed by homogenization in a stomacher for 10 min. The suspension was transferred into 50-ml Falcon tubes and centrifuged at 500 × g for 2 min. As a cryoprotectant, 20% (vol/vol) glycerol (Analar Normapur) (99.5%) was added to the supernatants. This mixture was gently blended, obtaining a fecal inoculum, which was then stored at −80°C.

Preculturing of the fecal inoculum before a 72-h in vitro colonic digestion was performed in BHI, unless mentioned otherwise (specialized media). Autoclaved penicillin flasks were filled with 36 ml of BHI broth and 4 ml of thawed fecal inoculum, flushed for 1 h with N₂ (1 bar), and incubated (Innova 42 [Incubator shaker series]; New Brunswick Scientific) at 37°C and 150 rpm for 24 h (referred to here as “FI₂₄”).

In a few digestion experiments, thawed glycerol stocks from specific strains (−80°C) were directly added to the rapeseed in vitro digestion (72 h) instead of first preculturing them (100 μl of stock in 40 ml of BHI).

(b) Stimulation of large intestine.

Autoclaved penicillin flasks were supplemented with rapeseed (1.6 g) using a 1/25 (s/liter) ratio. Subsequently, 40 ml of fluid consisting of a homogenous amount of FI₂₄ precultured bacteria and BHI broth (1:1 ratio) was added. The penicillin flasks were closed with Viton plugs, allowing the establishment of anaerobic conditions after 1 h of flushing with N₂. The flasks were then incubated for 72 h at 37°C and 150 rpm. Sampling occurred at the beginning and end of the incubation (0 h and 72 h). In some setups, samples were taken at other times specific to the experiment (24 h, 48 h, or FI₂₄) or before any incubation (FI_−80°C).

(iii) Digestion setups.

During a first digestion experiment, various porcine inocula (n = 20) were incubated with rapeseed in order to detect a suitable fecal inoculum producing a high level of TU and to gain insight into the distribution of the occurrence of fecal TU formation. In a second experiment, the fecal inoculum producing the highest level of TU was incubated both anaerobically and aerobically to evaluate the influence of oxygen on the in vitro digestion in BHI. In a third experiment, the influence of preculturing (24 h of culturing before 72 h of digestion), rapeseed addition, filter sterilization (0.22-μm-pore-size cellulose syringe filters; Whatman, Buckinghamshire, United Kingdom), and pasteurization (60°C for 0, 10, 30, and 60 min) of the precultured fecal inoculum FI₂₄ was investigated. Filter sterilization included a preliminary high-speed centrifugation step (20 min at 13,292 × g) to effectively remove fecal sludge prior to the use of the sterile filters, preventing leakage and clogging.

Bacterial identification. (i) Isolation of TU-producing bacteria.

The porcine inoculum with the highest TU production level was investigated for the presence of TU-producing bacterial groups by using five different media upon in vitro rapeseed digestion: BHI as general medium (control), SDx for fungi and yeasts, WCh for anaerobes, MacC for coliforms, and MRS for lactic acid bacteria (Oxoid). As an exception, 24-h preculturing of the fecal inoculum occurred in the respective specialized media instead of BHI, prior to the 72-h in vitro incubation with rapeseed (at 37°C and 150 rpm). This porcine inoculum was also subjected to in vitro digestion with and without rapeseed addition. Next, both in vitro digestions were plated out in five different dilutions (1/10 to 1/100,000) on BHI and MacC agar plates (based on previous results: see the “Contributing microorganisms” section below) with a spiral plater and incubated under aerobic (24 h) and anaerobic (for 48 h) conditions at 37°C. Based on colony morphology, different colonies from the rapeseed in vitro digestion were picked and reevaluated for TU formation upon rapeseed digestion. The bacterial digests producing TU were subsequently plated out in four dilutions (1/10 to 1/10,000) under the starting conditions (aerobic or anaerobic and BHI or MacC). Similarly, the morphologically divergent bacterial colonies were picked and evaluated through rapeseed in vitro digestions for TU formation. Also, these digests were plated out under the starting conditions but also in the opposite redox status (aerobic versus anaerobic) to evaluate facultative aerobic or anaerobic traits. The macroscopically different colonies were then picked and concurrently incubated in rapeseed in vitro digestions (TU detection) and subjected to streaking onto BHI agar medium to create monocultures. Once pure monocultures were achieved, bacterial colonies were picked for a last in vitro evaluation of TU formation (with and without rapeseed).

(ii) Biochemical and genetic identification.

Purified monocultures were phenotypically characterized by Gram staining and API 20E testing for preliminary identification (bioMérieux, Marcy l'Etoile, France). Isolates showing an observed TU concentration increase of more than 60%, when rapeseed was added versus when no rapeseed was added to the in vitro digestion (see Table 2), were selected for further identification tests.

TABLE 2.

Identification of thiouracil-producing bacteria^a

Identified bacterial species	% TU increase (with vs without RS)	Cell morphology	Gram stain	Oxidase	Catalase	Fermentation	Identification procedure(s) and result
							API 20E	Genetic
E. coli	88.1	Rods (0.7 by 1.0–1.8 μm), single, pairs, nonmotile	−	−	+	+	No ID	rpoB sequence analysis, DNA-DNA hybridization (ATCC 25290): Escherichia coli
E. coli	31.8	Rods	−	ND	ND	ND	E. coli 1 (98.8%)	ND
E. coli	90.6	rods (0.9 × 1.5–5.0 μm), single, pairs, nonmotile	−	−	+	+	E. coli 1 (88.3%)	rpoB sequence analysis, DNA-DNA hybridization (ATCC 25290): Escherichia coli
E. coli	26.5	Rods	−	ND	ND	ND	E. coli 1 (88.3%)	ND
S. enterica subsp. arizonae	62.4	Rods	−	ND	ND	ND	Salmonella enterica subsp. arizonae (99.1%)	ND
E. coli	89.6	Rods (0.9 by 1.5–10 μm), single, pairs, nonmotile	−	−	+	+	Enterobacter cloacae (38.8%)	rpoB sequence analysis, DNA-DNA hybridization (ATCC 25290): Escherichia coli
L. reuteri	93.7	Rods (1.0 by 1.5–3.0 μm), single, pairs, nonmotile	+	−	−	/	No ID	AFLP: Lactobacillus reuteri
L. reuteri	89.4	Rods (1.0 by 1.5–3.0 μm), single, pairs, nonmotile	+	−	−	/	No ID	AFLP: Lactobacillus reuteri
E. faecium	82.87	Cocci, 0.9-μm diam, single, pairs, short chains, nonmotile	+	−	−	/	Pantoea sp. 1 (48.9%)	AFLP: Enterococcus faecium

^aData represent percent thiouracil (% TU) increases between rapeseed (RS)-control digestions and morphological, biochemical, and genetic identification results. +, positive result; −, negative result; ND, not defined; No ID, no matching identification from database; AFLP, amplified fragment length polymorphism. Values in parentheses in column 8 represent percent identity.

For Gram-negative rods, the oxidative-fermentative (OF) test developed by Hugh and Leifson (36) was applied, and for those capable of fermentation of glucose (an indicator for Enterobacteriaceae), a partial rpoB sequence was determined, as such sequences allow species discrimination within this family (37). Total DNA was prepared according to the protocol of Niemann et al. (38). The rpoB gene was amplified and sequenced using the primers described by Mollet et al. (37). The PCR-amplified rpoB gene product was purified using a NucleoFast 96 PCR cleanup kit (Manchenery-Nagel, Düren, Germany). Sequencing reactions were performed using a BigDye Terminator sequencing kit (Applied Biosystems, Foster City, CA) and purified using a BigDye XTerminator purification kit (Applied Biosystems). Sequencing was performed using an ABI Prism 3130XL genetic analyzer (Applied Biosystems). Sequence assembly, the creation of a similarity matrix, and the phylogenetic analysis were all accomplished with the use of BioNumerics 5.1 software (Applied Maths, Sint-Martens-Latem, Belgium). The similarity matrix was based on a pairwise alignment using an open gap penalty of 100% and a unit gap penalty of 0%. The phylogenetic analysis, which included the consensus sequence in an alignment of rpoB sequences collected from the EMBL international nucleotide sequence library and from LMG reference strains, was performed afterward. A resulting tree was constructed using the neighbor-joining method.

For the isolates identified through partial rpoB sequencing, identification to the species level was not possible, and therefore DNA-DNA hybridizations were performed. Template DNA for this method was isolated according to a modification of the procedure of Wilson (39). Hybridizations were performed with DNA of a selected isolate and DNA of the type strain of Escherichia coli LMG 2092^T (BCCM/LMG Bacteria Collection, Ghent, Belgium), which was identified as the closest phylogenetic species of these isolates. Hybridizations were performed in the presence of 50% formamide at 43°C according to a modification (40, 41) of the method described by Ezaki et al. (42). Reciprocal reactions, i.e., the use of isolate DNA as a probe against DNA of LMG 2092^T (ATCC 25290) immobilized to a microplate well (A against B) and vice versa (B against A), were each performed in quadruplicate and the DNA-DNA relatedness was calculated from the mean value of A against B and that of B against A.

Isolates potentially belonging to the lactic acid bacteria (Gram positive and oxidase and catalase negative) were subjected to AFLP DNA fingerprinting (Keygene NV, Netherlands) for further identification. For this application, DNA was prepared using the method of Gevers et al. (43) with slight modifications. Amplified fragment length polymorphism (AFLP) DNA fingerprinting was performed as reported previously (44) using the restriction enzymes EcoRI and TaqI and the primer combination E01–6-carboxyfluorescein (6-FAM) and T11 (45). The obtained profiles were compared with reference profiles of lactic acid bacterial taxa (including Bifidobacteria) present in a BCCM/LMG in-house database using BioNumerics software. Clustering of the profiles was done using the Dice coefficient and the unweighted-pair group method using average linkages (UPGMA) algorithm.

Sample extraction and purification for TU detection.

Prior to the analysis of the digestion samples, sample preparation and cleanup were needed. These procedures have been described previously (9). Briefly, 5 ml of phosphate buffer (pH 8) was added to 1 ml of digest. The internal standard (PTU-D₅) was added at 50 μg kg⁻¹ followed by 100 μl of a methanolic derivative solution containing 3-iodobenzyl bromide (2 mg ml⁻¹). Upon 10 min of sonication, the derivatization was allowed to proceed in the dark at 40°C for 1 h. Afterwards, the pH of the reaction mixture was adjusted to 3.6 (±0.1) and different liquid-liquid extraction steps with 3, 2, and 2 ml of diethyl ether were applied. Once evaporated to dryness under a gentle N₂ stream (2 × 10⁵ Pa and 50°C), the samples were dissolved in 100 μl CH₂Cl₂ and 300 μl cyclohexane. Further sample cleanup consisted of solid-phase extraction with silica cartridges conditioned with cyclohexane and eluted with a 40/60 (vol/vol) mixture of hexane and ethyl acetate. Again, N₂ evaporation was applied. Samples were redissolved in 160 μl of 0.5% acetic acid and methanol (50/50).

Standard addition.

Quantification of TU in digestion samples was performed using the standard addition approach as described in Commission Decision 2002/657/EC (46). Each sample was divided over 2 vials with analogous mass (m) and volume (V) levels. One aliquot, the unknown, was added with a 50/50 mixture of 0.5% acetic acid and methanol (V_A), and the other, the known, was enriched with an equal amount of the analyte (TU) (V_known). The concentration of the addition solution had been determined previously by analyzing TU levels in digest samples and fitting these in a calibration curve in a digestion buffer. This estimation of the TU concentration was then spiked to the additional aliquot of each sample (ρ_A), giving rise to the area ratio of χ_known. Finally, quantification was established through evaluation of the area ratios of the known and unknown samples using the standard addition approach (21).

Quantification was established using the following formula: C_unknown = χ_unknown V_unknown ρ_A V_A/(χ_known V_known m_unknown − χ_unknown V_unknown m_known), with m_unknown = m_known, V_unknown = V_known, and C_unknown = χ_unknown ρ_A V_A/(χ_known − χ_unknown). In this formula, C and ρ are concentrations, χ is the area ratio, V is the volume, m is the mass, and A is the identified analyte (TU).

Instrumentation: LC-tandem mass spectrometry (LC-MS²).

Detection of TU was achieved with a liquid chromatograph (LC) coupled to a linear ion trap mass spectrometer (21). A Finnigan Surveyor LC system (Thermo Electron, San Jose, CA) was combined with a Symmetry C₁₈ column (Waters, Milford, MA) (5 μm by 150 mm by 2.1 mm) at 30°C running on a 50/50 solvent combination of 0.5% acetic acid (A) and methanol (B) at 0.3 ml min⁻¹. The linear gradient was passed off as follows for 35 min: A/B at 50/50 for 3 min, increasing to 0/100 for 17 min, and finally reequilibrating at 50/50 for 10 min. The linear LTQ ion trap mass spectrometer (Thermo Electron, San Jose, CA) was fitted with a heated electrospray ionization probe (HESI) operating in the negative-ion mode. Applied working conditions were as follows: source voltage at 5 kV; capillary voltage at −50 V; tube lens voltage at −128.04 V; vaporizer and capillary temperatures at 250°C and 275°C; and sheath and auxiliary gas at 30 and 5 arbitrary units (a.u.), respectively. Based on its good in-house performance in urine and a secondary validation in a digestion suspension, confirming these performance criteria (with a decision limit [CC_α] of 1.63 ng TU g⁻¹ rapeseed, a mean percentage of recovery of 102.73 ± 10.95, and a repeatability percent relative standard deviation [RSD%] of 10.66), the method was found adequate for screening of TU in digests. Measured transitions are reported in Table 1.

TABLE 1.

The transitions monitored by LC-MS² for 2-thiouracil and its internal standard, 6-propyl-D₅-2-thiouracil

Analyte	[M-H]−	Product ions	Collision energy (eV)
2-Thiouracil	343	182, 215, 309	44
6-Propyl-D5-2-thiouracil	390	127, 262, 356	30

Quality assurance.

Preceding the LC-MS² analysis, a standard mixture of the target compound and the internal standard was injected to check the operational conditions of the LC-MS² device. Identification of TU was based on the retention time relative to that of the internal standard and on the ratios of the product ions according to Commission Decision 2002/657/EC (46). Every in vitro digestion simulation was conducted in 3-fold, and the internal standard was added prior to cleanup (n = 3).

Data handling.

All data processing was performed with XCalibur 2.0.7 (Thermo Fisher Scientific, San Jose, CA). Statistical testing (Student's t test) was carried out with SPSS 21 (IBM, Belgium) to assess the significance (P ≤ 0.05) of the TU yields. Normality and equal variances were set as boundary conditions.

RESULTS

Porcine intestinal thiouracil formation.

TU detection levels resulting from 20 fecal porcine in vitro digestions with rapeseed ranged between 33.4 (±0.1) and 80.5 (±2.8) ng g⁻¹ rapeseed for sows (S1 to S10) and between 3.3 (±0.8) and 16.5 (±3.4) ng g⁻¹ rapeseed for gilts (G1 to G10) (Fig. 1). The gilts were female porcine livestock which were about to be inseminated for the first time, whereas the sows were multiparous adult females. Statistical differences (P = <0.001) between the two life stages (sow and gilt) regarding their TU production upon rapeseed digestion were proven.

FIG 1.

Results of thiouracil (TU) production in 10 gilts (G1 to G10) and 10 sows (S1 to 10) during in vitro digestive simulations with rapeseed (n = 3) for 72 h (ng TU g⁻¹ rapeseed ± SD).

Furthermore, anaerobic and aerobic conditions were compared for rapeseed-based TU production. TU production under anaerobic conditions was significantly (P = 0.015) higher than TU production under aerobic digestion conditions at the end of the incubation (T₄₈), namely, 55.04 ± 4.73 ng g⁻¹ rapeseed and 17.70 ± 8.86 ng g⁻¹ rapeseed, respectively.

In vitro colon digestion and thiouracil formation: influential factors.

To explore the mechanism behind TU formation, a porcine inoculum producing a high level of TU was exposed to either various periods (0, 10, 30, and 60 min) of pasteurization (60°C) or filter sterilization (0.22-μm-pore-size filter) or a combination of the two procedures prior to a rapeseed in vitro digestive simulation. No significant increase in TU formation could be demonstrated after 72 h of digestion upon filter sterilization in any combination with pasteurization (Fig. 2). Pasteurization alone did not affect the TU-forming potential of the fecal inoculum since a significant (P = 0.048) TU increase similar to that of the control (P = 0.036) could be seen at up to 30 min into pasteurization (60°C). Beyond this duration, variations in TU formation levels increased, but no significant difference in TU formation results was noticed (Fig. 2).

FIG 2.

Effect of filter sterilization (FS) and pasteurization (0, 10, 30, and 60 min at 60°C) of the precultured fecal porcine inoculum (FI₂₄) on thiouracil formation of a high-producing porcine fecal inoculum during in vitro digestive simulation with rapeseed (ng TU g⁻¹ rapeseed ± SD) (n = 3) at the start (T₀) and toward the end (T₇₂) of the incubation.

To identify the origin of the background TU levels prior to digestive simulation, in vitro incubations of BHI broth as such, a non-a priori-cultivated fecal inoculum (FI_−80°C), FI₂₄, rapeseed (RS), and autoclaved rapeseed (autocl. RS) in specific combinations were performed (Fig. 3). Trace levels of TU in thawed fecal inoculum (T_−80°C) prior to any incubation and in incubated (T₇₂) BHI medium were observed. After 24 h (FI₂₄) of preculturing of the fecal inoculum (FI), a significant (P = <0.05) increase in the TU concentration could be detected compared to T₀. After 72 h of incubation, the in vitro digestions of FI₂₄ and BHI without rapeseed produced amounts of TU similar to those seen with the same digestions with rapeseed and with autoclaved rapeseed, respectively. The digestions that did not contain FI₂₄ showed no significant increase of TU levels over time, except for BHI with rapeseed, which showed a significant (P = 0.027) but minor increase of TU compared to the T₀ results.

FIG 3.

Overview of the effects on thiouracil formation of rapeseed (RS), autoclaved rapeseed (AUTOCL.RS), and brain heart infusion (BHI) by a whole porcine fecal inoculum at the start (T₀) and toward the end (T₇₂) of the incubation. In addition, thawed fecal inoculum (FI_−80°C) and precultured fecal inoculum (FI₂₄) were evaluated for thiouracil formation (ng TU ml⁻¹ ± SD) (n = 3).

Bacterial isolation and identification. (i) Contributing microorganisms.

For preliminary identification of bacterial groups potentially contributing to TU formation upon Brassicaceae digestion, the porcine fecal inoculum with the highest TU production level was incubated in one nonselective BHI and four group-specific broths. TU production was the highest in MacC (39.5% ± 1.9% TU) and MRS (25.8% ± 2.8% TU) media relative to the nonselective BHI (100% TU) (Fig. 4).

FIG 4.

Overview of thiouracil production (%) by a porcine fecal inoculum producing a high level of thiouracil in different broths (MacC, coliforms; MRS, lactic acid bacteria; SDx, anaerobes; WCh, fungi and yeasts) after preculturing prior to incubation (T₀) and after 72 h of incubation with rapeseed (T₇₂) (percent TU abundance ± SD) (n = 3).

(ii) Isolation and identification.

To identify and isolate TU-producing bacterial species, the intestinal inoculum of a sow culture producing the highest level of TU was selected for in vitro digestion, with and without rapeseed addition. Finally, 13 fecal suspensions were found to be positive for TU after this process, from which only 8 digestions with sufficient TU production ranging from 18.47 to 31.88 ng g⁻¹ rapeseed were selected.

Next, 12 pure monocultures were achieved and investigated using API 20E. Only four of these isolates could be identified directly based on the APIweb database: three as Escherichia coli (percent identification [ID%], 98.0, 88.3, and 88.3) and one as Salmonella enterica subsp. arizonae (ID%, 99.1). Of the remaining isolates, three were not further investigated, because they were not producing large amounts of TU. The remaining five isolates showed an observed TU concentration increase of more than 80% when rapeseed was added to the in vitro digestion in comparison to when none was added (Table 2). These isolates and an E. coli representative identified previously by API (90.6% TU [TU%] increase) were selected for further identification.

Three isolates were identified as Escherichia spp. that were probably E. coli isolates as determined on the basis of rpoB gene sequence analyses, while three others were identified as lactic acid bacteria (one as Enterococcus faecium and two as Lactobacillus reuteri) on the basis of AFLP DNA fingerprinting. One representative Escherichia isolate was additionally investigated through DNA-DNA hybridization. DNA of this isolate was compared with DNA of the type strain of E. coli [LMG 2092^T]), and 75% DNA-DNA relatedness was found, which is above 70%, the value generally regarded as the limit of species delineation (47). Therefore, the Escherichia isolates were confirmed to belong to E. coli. In total, nine isolates were identified at the species level through the different identification procedures (Table 2). Their respective TU increases upon rapeseed in vitro incubation are presented in Fig. 5.

FIG 5.

Recovered thiouracil concentrations (ng g⁻¹ rapeseed ± SD) during precultured in vitro incubations with rapeseed (n = 3) of the 9 identified TU-producing isolates (Enterococcus faecium, Escherichia coli [a to e], Lactobacillus reuteri [a and b], and Salmonella enterica subsp. arizonae) at the start (T₀) and toward the end (T₇₂) of the incubation.

To elucidate precursors of microbial TU formation in the substrate, Lactobacillus reuteri and Enterococcus faecium, as representatives of isolates producing high and low TU levels (Fig. 5), were exposed to various preparations (rapeseed, autoclaved rapeseed, and no rapeseed). Similar patterns were observed for the two bacterial species, with TU concentrations when rapeseed was added that were significantly (P ≤ 0.05) higher than those seen with autoclaved rapeseed and control (no rapeseed) incubations (Fig. 6). TU concentrations in autoclaved rapeseed incubations were not significantly (P > 0.05) different from those seen with the control incubations.

FIG 6.

Influence of rapeseed on thiouracil production (ng g⁻¹ ± SD) upon in vitro incubation (n = 3) with purified Lactobacillus reuteri and Enterococcus faecium (not precultured). Autocl., autoclaved.

Lactobacillus reuteri, Enterococcus faecium, and also Escherichia coli were then plated onto BHI agar and onto the more selective agars, MRS (Lactobacillus reuteri and Enterococcus faecium) and MAcC (Escherichia coli), for enumeration after incubation with and without rapeseed, without preculturing the organisms to evaluate whether the bacteria developed sufficiently over time and whether the presence of rapeseed influenced bacterial growth. Results showed that the log₁₀ value ranged between 7 and 9 ml⁻¹ at T₇₂. No significant difference between rapeseed and control incubations was found regarding bacterial counts, except for L. reuteri, which showed a significant (P ≤ 0.05) positive increase in the bacterial count in the rapeseed digestion (8.23 ± 0.35 log₁₀ ml⁻¹) compared to the control digestion (7.11 ± 0.35 log₁₀ ml⁻¹).

DISCUSSION

In the present study, 9 cultures belonging to four different bacterial species, E. coli, E. faecium, L. reuteri, and S. enterica subsp. arizonae, were isolated from porcine feces and identified as involved in the production of the banned thyreostat thiouracil upon rapeseed incubation. The mechanism bringing about endogenous thiouracil production was further investigated by altering digestion parameters, providing strong evidence that an extracellular membrane-bound (myrosinase-like) bacterial enzyme formed under anaerobic conditions and the addition of rapeseed were pivotal in this process.

The levels of microbial intestinal TU formation in vitro observed in the present study were significantly different between porcine life stages. It has indeed been previously reported that the diversity of the intestinal microbiota in pigs increases with age (48, 49, 50), which could explain the higher levels of TU production by fecal inocula derived from elder sows. Moreover, the levels of E. coli and Lactobacillus sp. have been previously described as being lower in young pigs (48, 50). For E. coli specifically, Katouli et al. (51) showed that the mean number of biochemical phenotypes in piglets increased as animals aged and that sows normally had a higher number of biochemical phenotypes than their offspring. They subsequently reported that the dominant E. coli strains in piglets are quite different from those in their sows. Therefore, the number of different intestinal bacterial strains, as well as the present dominant types of a strain, might potentially be important for intestinal TU formation (51). These findings suggest that colonic bacterial fermentation significantly contributes to the prevalence of endogenous thiouracil in livestock urine. Of course, relating these findings to in vivo data remains difficult. It is, however, likely that naturally formed thiouracil residues are excreted in urine upon absorption through the intestinal tract, a process which has been previously observed in vivo with TU concentrations below 10 ppb upon Brassicaceae feeding in bovines (1). Orally administered thyreostatic drugs have been reported to rapidly occur in urine and reach a cumulative total of 6.49% to 16.7% of the administered dose within 24 h in rats (52). Whether this excretion rate based on oral administration is also applicable to the excretion of intestinally formed TU remains to be investigated.

Notwithstanding, the detection of TU in urine samples of livestock is a widespread phenomenon, but urinary residues are not consequently recorded in every animal. A survey study based on national control plan data in France found 32% TU-positive samples in bovines (n = 1,089) and 80% in porcine livestock (n = 201) (53). That study showed that concentrations of TU were higher in bovines than in porcine livestock, considering only porcine livestock between 1 month and 8 months old, which from our results show significantly lower TU-producing potential in vitro than older pigs. In a Polish survey, only 14% of bovines (n = 279) and 13% of pigs (n = 258) tested positive for TU (54), but no age categories were mentioned in that study. In this context, the influence of age, gender, and, in a minor way, country on the determination of a potential TU threshold value in vivo has been previously described (53, 54).

To propose a mechanistic explanation for TU formation upon Brassicaceae digestion, a porcine intestinal inoculum producing TU at a high level was evaluated under various in vitro conditions. The use of various selective media pointed toward the involvement of coliform and lactic acid bacteria, which are abundantly present in the gut microbial communities of pigs (29, 55). Moreover, further elucidation of the influence of redox conditions reaffirmed the fact that anaerobic microorganisms outnumber aerobic bacteria by a factor of at least 3 to 5 log₁₀ (49) in the intestinal environment of pigs. This explains the lower TU concentration monitored under aerobic conditions. Filter sterilization showed no increase of the TU concentration upon in vitro digestion, providing evidence for the requirement of (active) bacterial cells for TU formation to take place. Background TU traces present prior to incubation may be attributed to the preculturing of bacteria in the presence of other fecal constituents. During this period, the fecal inoculum can metabolize any feed-related remains, which are inherently present in the fecal suspension, consequently resulting in the production of TU. Indeed, when no preculturing was applied, TU formation was marginal. Pasteurization of the precultured inoculum did not significantly affect TU formation. It has been found that E. coli can survive at 60°C for up to 30 min depending on the strain and the surrounding matrix (56). At 55°C, inactivation is achieved only after 60 to 120 min in broth (56). Although heat-treated microorganisms generally do not grow on agar, they can succeed in repairing damage and regain their potential for growth under optimal broth conditions (56). Such events might explain the undisturbed TU production over the first 30 min of pasteurization, while bacterial survival after 60 min is unlikely due to irreversible damage. Assuming that the responsible (bacterial) mediator is a membrane-bound protein (myrosinase-like enzyme), heat treatment would likely cause some degree of denaturation over time, but semireversible denaturation has been reported (57). Moreover, β-thioglucoside glucohydrolases remain active in a broad range of temperatures (up to 60°C) and pHs (pH 5 to 10) (58), offering an explanation for sustained TU production upon pasteurization and supporting the potential hypothesis of extracellular membrane-bound myrosinase-like enzyme as the mediating factor for TU formation.

The identified bacterial isolates with TU-forming ability include some of the most prevalent intestinal bacteria, i.e., lactobacilli, coliforms, and enterococci populating the intestine of pigs (49, 59). Since these bacteria are common residents of the intestine in many animal species, the occurrence of TU in urine of several animal species, as reported earlier (2), may be easily explained. Enterococcus faecium is known as the most prevalent Enterococcus sp. in porcine livestock (60) and has also been found to metabolize glucosinolates in vitro (61). Both Lactobacillus sp. and Enterococcus sp. have been demonstrated to express β-thioglucosidase activity (62). In a study with rats transfaunated with Lactobacillus sp. (LEM 220) and fed with rapeseed meal, an increase in goitrogenicity was observed compared to the results seen with germfree rats (26, 63). Similar results were obtained in gnotobiotic E. coli (EM0) mice fed rapeseed meal (64). Moreover, degradation of progoitrine, a precursor of goitrine (a natural thyeostat), by E. coli (65) and of the glucosinolate sinigrin by L. agilis (R16) (27) have also been described. Specifically in rodents, pigs, and chickens, L. reuteri is one of the dominant species in the gastrointestinal tract (66). This lactic acid bacterium has the capacity to form reuterin, an antimicrobial agent, and has therefore been proposed as a potential additive for pelleted feeds in pigs (67). Two of the prerequisite characteristics for pelleting were resistance to 70°C for 10 s and resistance to gastric and intestinal pHs. Previously known as S. choleraesuis subsp. arizonae, S. enterica subsp. arizonae has been isolated from reptiles, fowl, turkeys, ducks, dogs, cats, monkeys, goats, wild boars, and, lately, pigs (68), but no glucosinolate-degrading or myrosinase-like capacities have been described yet for this taxon.

To confirm that Brassicaceae feed was the source for TU formation, L. reuteri and E. faecium as bacteria producing high and low levels of TU, respectively, and a porcine fecal inoculum producing a high level of TU were selected for incubations. Autoclaved rapeseed and control incubations with pure strains did not differ significantly in TU concentrations. Addition of rapeseed as a whole did, however, significantly increase TU formation upon incubation with pure strains, reaffirming its importance as a source of TU precursors, in line with previous in vitro (28) and in vivo (1) reports. During in vitro digestion with the porcine fecal inoculum producing TU at a high level, levels of TU formation did not significantly differ between autoclaved and nonautoclaved rapeseed administrations. This may be explained by the prior assumption that fecal constituents present in the inoculum can act as precursors for TU formation when no rapeseed is added. Indeed, it has been reported in the literature that 60% of the glucosinolate content of feed reaches the colon unaltered in pigs and that, although the content is mostly degraded prior to defecation, many breakdown products are present in feces (69). The autoclaved rapeseed and control in vitro digestions produced equal amounts of TU upon incubation with fecal inoculum and pure strains. Autoclaved and, thus, sterile and possibly denatured rapeseed may not contain the required precursors and as a result may produce results that do not differ significantly from those seen in the absence of rapeseed. Oerlemans et al. (70) reported that substantial breakdown of all glucosinolate groups at temperatures above 110°C was observed, which may also be the case during autoclave processing (121°C). Besides, it is possible that the rich BHI medium, which consists of animal tissues (brain and heart), is additionally responsible for low TU levels in bacterial incubations without rapeseed or autoclaved rapeseed, since it has been reported that 2-thiouracil can be found in tRNA of E. coli and that N-glucosidases from E. coli extracts are able to cleave uracil free from DNA, which is one of its building blocks (71).

Finally, the influence of the substrate (rapeseed) on bacterial growth was evaluated through enumeration of some of the pure strains to ensure that increased bacterial growth was responsible for enhanced TU formation and not the feed itself. Indeed, it has been reported that certain hydrolysis products from various Brassica crops exert inhibiting influences in vitro on bacterial growth of Enterobacteriaceae and certain enterococci in pigs (72). In the present study, however, no negative effects on bacterial growth were observed as a result of rapeseed administration for E. coli and E. faecium. In contrast, minor growth stimulation was observed for L. reuteri when rapeseed was added.

The absence of an effect of rapeseed on bacterial growth and its restricted growth stimulation in the case of L. reuteri may be explained by the variety of glucosinolates present in rapeseed, which are mainly of aliphatic origin. Saavedra et al. (72) investigated the effect of glucosinolate hydrolysis products on intestinal bacteria. Indolic and aromatic glucosinolates, but also two aliphatics (sinigrin and glucoraphanin), were mainly considered. Depending on the type of hydrolysis product and its concentration as well as the type of bacteria exposed to it, the inhibiting effect can be hampered. From these results, it may be concluded that TU formation is directly related to the presence of rapeseed feed and its inherent glucosinolates and/or degradation products thereof, but other sources of TU precursors may not be excluded.

In summary, fecal inocula of porcine livestock in two different physiological stages of life were screened for their capability to produce the banned thyreostat thiouracil upon rapeseed digestion, resulting in the observation of a microbial capacity to produce TU that increased with age. Furthermore, 9 TU-forming bacterial species were isolated from a sow producing high levels of TU and identified as belonging to the species Escherichia coli, Enterococcus faecium, Lactobacillus reuteri, and Salmonella enterica subsp. arizonae. Moreover, the mechanism behind this bacterially mediated formation was tentatively characterized, demonstrating that a cell-dependent, membrane-bound enzyme that was moderately heat resistant (tolerating 60°C for up to 30 min) was involved. In addition, it may be concluded that rapeseed as a feed can be considered a possible source of TU precursors, sustaining the hypothesis that glucosinolates are subjected to bacterial myrosinase-like enzyme activity on the basis of this observation. These new insights into endogenous TU formation in livestock offer a mechanistic explanation for observations of low-level TU formation but predominantly reinforce the need for more specifically tailored European Union legislation.