The discovery of the adventitious formation of the potential cancer-causing agent acrylamide in a variety of foods during cooking has raised much concern, but the chemical mechanism(s) governing its production are unclear. Here we show that acrylamide can be released by the thermal treatment of certain amino acids (asparagine, for example), particularly in combination with reducing sugars, and of early Maillard reaction products (N-glycosides). Our findings indicate that the Maillard-driven generation of flavour and colour in thermally processed foods can -- under particular conditions -- be linked to the formation of acrylamide.

The dietary mutagens 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) are activated to genotoxins by rat and human liver cytochrome P450 (P450) 1A1- and 1A2-mediated N-oxidation. Immunoquantitation of 51 human liver samples revealed a wide range in P450 1A2 expression (10-250 pmol/mg of microsomal protein, median 71 pmol/mg), with 39% of the livers containing >100 pmol/mg of protein. There was no evidence for expression of P450 1A1 (<1 pmol/mg of protein). P450 1A2 levels were correlated to MeIQx and PhIP N-oxidation rates (r = 0.83, 0.73, respectively). In male Fischer-344 and Sprague-Dawley rats, hepatic P450 1A2 ranged from 5 to 35 pmol/mg of protein, while P450 1A1 was <1 pmol/mg. Animal pretreatment with 3-methylcholanthrene, beta-naphthoflavone, or polychlorinated biphenyls (PCB) resulted inasmuch as 340-fold and >1000-fold induction of P450 1A2 and 1A1, respectively, and a 220-fold increase in N-oxidation activity. Approximately 20% of the human samples were as active in N-oxidation and conversion of MeIQx to bacterial mutagens as microsomes of PCB-pretreated rats [3-4 nmol of NHOH-MeIQx formed min-1 (mg of protein)-1]. In contrast, microsomes from PCB-treated rats displayed higher rates of PhIP N-oxidation and activation to mutagens than the most active human liver microsomes [8-24 vs 2-4 nmol of HNOH-PhIP formed min-1 (mg of protein)-1]. Recombinant human P450 1A2 showed catalytic efficiencies of MeIQx and PhIP N-oxidation that were 10-19-fold higher than purified rat P450 1A2. Cytochrome P450 1A2 expression in rodent and human liver tissue varies greatly and there are considerable differences between the enzymes in the two species in the activation of some heterocyclic aromatic amines, which must be considered when assessing human health risk.

The formation of acrylamide was studied in low-moisture Maillard model systems (180 degrees C, 5 min) based on asparagine, reducing sugars, Maillard intermediates, and sugar degradation products. We show evidence that certain glycoconjugates play a major role in acrylamide formation. The N-glycosyl of asparagine generated about 2.4 mmol/mol acrylamide, compared to 0.1-0.2 mmol/mol obtained with alpha-dicarbonyls and the Amadori compound of asparagine. 3-Hydroxypropanamide, the Strecker alcohol of asparagine, generated only low amounts of acrylamide ( approximately 0.23 mmol/mol), while hydroxyacetone increased the acrylamide yields to more than 4 mmol/mol, indicating that alpha-hydroxy carbonyls are much more efficient than alpha-dicarbonyls in converting asparagine into acrylamide. The experimental results are consistent with the reaction mechanism based on (i) a Strecker type degradation of the Schiff base leading to azomethine ylides, followed by (ii) a beta-elimination reaction of the decarboxylated Amadori compound to afford acrylamide. The beta-position on both sides of the nitrogen atom is crucial. Rearrangement of the azomethine ylide to the decarboxylated Amadori compound is the key step, which is favored if the carbonyl moiety contains a hydroxyl group in beta-position to the nitrogen atom. The beta-elimination step in the amino acid moiety was demonstrated by reacting under low moisture conditions decarboxylated model Amadori compounds obtained by synthesis. The corresponding vinylogous compounds were only generated if a beta-proton was available, for example, styrene from the decarboxylated Amadori compound of phenylalanine. Therefore, it is suggested that this thermal pathway may be common to other amino acids, resulting under certain conditions in their respective vinylogous reaction products.

3-Mono-chloropropane-1,2-diol (3-MCPD) is a contaminant that occurs in food in its free (diol) form as well as in an esterified (with fatty acids) form. Using a simple intestinal model, it was demonstrated that 3-MCPD monoesters and 3-MCPD diesters are accepted by intestinal lipase as substrates in vitro. Under the chosen conditions, the yield of 3-MCPD from a 3-MCPD monoester was greater than 95% in approximately 1 min. Release from the diesters was slower, reaching about 45, 65 and 95% of 3-MCPD after 1, 5 and 90 min of incubation, respectively. However, in human, the hydrolysis of 3-MCPD esters is unlikely to release 100% as 3-MCPD, as triglycerides and phospholipids are hydrolysed in the intestine liberating 2-monoglycerides. Assuming a similar metabolism for 3-MCPD esters as that known for acylglycerols in humans in vivo, the de-esterification in positions 1 and 3 would thus be favoured by pancreatic lipases. Therefore, 3-MCPD, and 3-MCPD-2 monoesters would be released, respectively, from the 1-/3-monoesters, and the diesters potentially present in food. Hence, information on the exact amounts of the partial fatty acid chloroesters, i.e. 3-MCPD mono- and diesters, is important to assess the contribution of foods to the bioavailability of 3-MCPD. Therefore, a rapid method for the determination of the ratio of 3-MCPD monoesters to diesters in fats and oils was developed using gas chromatography-mass spectrometry (GC-MS) and isotopically labelled 3-MCPD esters as internal standards. The analysis of 11 different samples of fat mixes typically employed in food manufacturing demonstrated that a maximum of about 15% of the total amount of 3-MCPD bound in esters is present in the monoesterified form. The potentially slower release of 3-MCPD from 3-MCPD diesters, and the mono- to diesters ratio suggest that 3-MCPD esters may in fact contribute only marginally to the overall dietary exposure to 3-MCPD. Further work on the bioavailability, metabolism and possible toxicity of chloroesters per se is warranted.