Molecular Toxicology and Clinical Pharmacology, Imperial College
School of Medicine, London, United Kingdom
The cooking of meat has been found to generate compounds that
possess extreme mutagenicity when examined in short term tests. This
observation led to the isolation and identification of a family of
mutagenic chemicals, all of which are heterocyclic amines. These amines
are potent bacterial and eukaryotic cell mutagens, and all of those
tested have been found to induce tumors in laboratory animals.
Metabolic activation of the heterocyclic amines predominantly involves
CYP1-mediated N-hydroxylation and then
O-esterification by phase II enzymes. In contrast, carbon
oxidation, glucuronidation, and sulfation reactions at sites other than
the hydroxylamine yield detoxication metabolites. In humans, the
activities of these pathways are known to vary between individuals and
are likely to influence susceptibility to the genetic toxicity of the
heterocyclic amines. Clearly, accurate determination of human exposure
to the heterocyclic amines and identification of the key enzyme systems involved and their regulation will be required for rational assessment of the risk and will help devise strategies to reduce such risk.
 |
Introduction |
Epidemiological
studies show that the incidence of human cancer is associated with
lifestyle, occupation, iatrogenic, and cryptogenic factors (Williams,
1985
). By far the most significant of these is lifestyle, particularly
diet and tobacco use, which together may account for ~70% of all
human cancer (Doll and Peto, 1981). Among dietary factors, high fat
intake, low fiber intake, and the consumption of well cooked meat seem
to predispose individuals to cancer of the colon, breast, pancreas,
prostate, and the endometrium. The discovery that cooked food can be
mutagenic (Commoner et al., 1978) led to efforts to isolate the
mutagenic products. Groups in Japan and America showed that the cooking
of food generates many compounds that are mutagenic in short-term
bacterial assays including polycyclic aromatic hydrocarbons,
nitrosamines, and a family of compounds that were all heterocyclic
amines (HAs1) (Sugimura and Sato, 1983; Felton et
al., 1986). While polycyclic aromatic hydrocarbons were shown to be the
major mutagens on a mass basis, the HAs were found to be by far the
more potent mutagens. These latter compounds include
imidazoquinoxalines, imidazoquinolines, and imidazopyridines (see Fig.
1) and are found when food, particularly red meat, is cooked under normal household conditions; they are not
present in uncooked food.
| |
Human Exposure to the Heterocyclic Amines |
Formation of the HAs has been investigated in model systems, and a
requirement for sugars, amino acids, and creatinine (or creatine) has
been established (Jagerstad et al., 1991; Skog et al., 1992). These
naturally occurring compounds are all present in red meat and can react
together during cooking in Maillard reactions, through which food
acquires its characteristic flavors, odors, and appearance. By-products
of this chemistry include the HAs (Jagerstad et al., 1983). Three
of these cooked food-derived HAs,
2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx),
2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline (DiMeIQx), and
2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP), are found
more frequently than the others; together they account for the majority
of the genotoxic potential of fried beef (Felton et al., 1986; Murray
et al., 1988; Murray et al., 1993). These three compounds have been
detected in beef, lamb, pork, chicken, and fish, especially when
chargrilled, fried, or roasted. HAs are present in microgram per gram
quantities in cooked meat; nevertheless, human exposure to the HAs,
although low, is chronic since meat is consumed daily over a lifetime.
Thus, depending upon dietary preferences, an individual's daily
exposure to heterocyclic amines is likely to range from microgram
quantities to essentially zero in the case of vegetarians.
Significantly, the incidence of cancer is approximately 60% lower in
vegetarians than in nonvegetarians (Thorogood et al., 1994).
In studies of the fate of administered HAs, it was found that
irrespective of the route of administration, oral, intraperitoneal, or
intravenous, similar amounts of administered HA radiolabel were
excreted in the urine and the feces (Gooderham et al., 1987). This and
the fact that most of the radioactivity was eliminated in the first
24 h indicated that the HAs were rapidly and extensively absorbed
(Gooderham et al., 1987, 1991). Analysis of the excreted material
showed that the majority of an administered dose of heterocyclic amine
was biotransformed but that a small proportion (2-5%) was excreted in
the urine as unchanged parent amine. Subsequently, similar results were
obtained in studies of human volunteers fed fried beef meals (Murray et
al., 1989; Lynch et al., 1992) (see Fig.
2). In one such study, fried beef
containing known quantities of HAs was fed to healthy volunteers on
each of four separate occasions, and urinary excretion of unchanged HAs
was examined. Irrespective of the dose, about 2% of the ingested MeIQx
and between 0.5 to 1% of the PhIP was excreted unchanged in the urine
within 24 h of the test meal, the majority being eliminated within
8 h (Murray et al., 1989; Lynch et al., 1992). Although there was up to 5-fold interindividual variation in the excretion of unchanged amine (Fig. 2), within an individual the percentage excreted on different occasions remained remarkably consistent (Lynch et al., 1992). Excretion of unchanged amine in urine is a function of the
extent of absorption and clearance due to metabolism. From animal
studies, HA absorption seems to be almost complete; thus, variations in
urinary excretion should provide a measure of interindividual variability in the metabolism of these compounds.
| |
Metabolism of the Heterocyclic Amines |
The rapid elimination of HA radiolabel and that only a small
amount of amine was excreted unchanged in the urine suggested that
these compounds are likely to be extensively metabolized (Gooderham et
al., 1987, 1991). In vitro studies with MeIQx and PhIP using liver
microsomal preparations from rats, mice, and rabbits showed that at
least two oxidative metabolites were formed from each compound, a
ring-hydroxylated product and the N-hydroxy derivative
(Gooderham et al., 1987; Turesky et al., 1988, 1991; Sjodin et al.,
1989; Turteltaub et al., 1989; Watkins et al., 1991a,b). In addition,
both the parent amines and their primary oxidative metabolites can be
further biotransformed to a variety of phase II metabolites including
glucuronides (Kaderlik et al., 1994), sulfate esters (Chou et al.,
1995), and acetylated products (Lin et al., 1995). As an example, the
routes of PhIP metabolism are shown in Fig.
3. Examination of the primary oxidative
metabolites in a mutagenicity assay such as the Ames Salmonella
typhimurium test showed that the N-hydroxy metabolites
of MeIQx and PhIP were direct-acting mutagens, whereas the
ring-hydroxylated products were not (Rich et al., 1992; Zhao et al.,
1994). Analysis of HA metabolism by human liver microsomal fractions
showed N-hydroxylation to be the primary oxidative route of
metabolism of these HAs with Km values of
60 and 55 µM for MeIQx and PhIP, respectively, with little if any
aromatic hydroxylation (Rich et al., 1992; Zhao et al., 1994). Clearly,
there are species differences in the oxidative metabolism of HAs since
experimental animals are able to both activate and detoxify these
amines, whereas humans convert them predominantly to their reactive
genotoxic metabolite.
Studies using a variety of different approaches have shown that the
genotoxic N-hydroxylation pathway of these amines involves primarily members of the CYP1A subfamily (Watanabe et al., 1982; McManus et al., 1989; Turteltaub et al., 1989; Turesky et al., 1991;
Watkins et al., 1991a,b; Rich et al., 1992; Zhao et al., 1994). With
human liver microsomes, the highly specific and sensitive inhibitor of
human CYP1A2, furafylline, inhibited the formation of
N-hydroxy MeIQx by >90% and of N-hydroxy PhIP
by about 85% (Table 1). These in vitro
findings were confirmed in vivo in a double blind, crossover study
(Boobis et al., 1994). Six adult male volunteers were asked to refrain
from eating meat or meat products for 24 h before a test meal. On
the day of the study they ingested placebo or furafylline (125 mg) and
2 h later consumed a fried beef meal containing a known amount of
HAs (determined by gas chromatography/mass spectrometry). In the
case of MeIQx, subjects on placebo excreted about 2.2% of the ingested
dose as unchanged amine in the urine, but after furafylline this
increased almost 15-fold (Table 1). In the case of PhIP, the placebo
leg subjects excreted about 1.2% of the administered dose as unchanged PhIP in the urine, whereas after furafylline, this increased about 3.5-fold (Table 1). Both the absolute amount of unchanged amine and the
time over which it was excreted were increased after furafylline, demonstrating the substantial contribution of CYP1A2 to the metabolism of these amines in humans.
There are reports that CYP1A1, CYP1B1, CYP3A4, CYP2C9, and CYP2A3 are
all capable of metabolizing HAs to their genotoxic N-hydroxy derivatives (McManus et al., 1989; Yamazaki et al., 1993; Edwards et
al., 1994; Crofts et al., 1997; Hellmold et al., 1998). However, all of
these P450 isoforms are less active toward HA substrates than
CYP1A2. An illustrative comparative study has been reported by Crofts
et al. (1998) in which human P450 isoforms were expressed in insect
cells and the kinetic constants of PhIP N-hydroxylation determined under comparable conditions (Table
2). CYP1A1 and CYP1A2 had comparable
N-hydroxylation activity, while CYP1B1 was much less active.
CYP1A1 had the highest 4-hydroxylation activity. Levels of CYP1A2 in
human liver can vary considerably (Sesardic et al., 1990); thus,
hepatic metabolism of heterocyclic amines such as PhIP will vary within
the human population. CYP1A2 expression is almost exclusively hepatic,
whereas CYP1A1 and CYP1B1 have been detected in a variety of
extrahepatic organs, usually after exposure to inducing agents. Hence,
the hepatic oxidative metabolism of the HAs will be CYP1A2-dependent,
whereas in extrahepatic tissues metabolism is likely to be supported by
CYP1A1 and to a lesser extent by CYP1B1. Again, variations in the
expression of these extrahepatic enzymes will contribute to variation
in the overall disposition and toxicity of these compounds.
It has recently been shown that human urinary metabolites of the HAs
MeIQx and PhIP include glucuronides and sulfate esters (Turesky et al.,
1998; Malfatti et al., 1999). Indeed,
N-hydroxy-PhIP-N2-glucuronide
is thought to be the major urinary metabolite of PhIP accounting for
about 50% of the dose (Malfatti et al., 1999). At least five
glucuronides of PhIP have been reported, the
N2-glucuronide,
N2-hydroxy glucuronide, the
N3-glucuronide, the
N3-hydroxy glucuronide, and the
4'-hydroxy glucuronide. Since the glucuronides can be eliminated via
bile, it is possible that intestinal bacterial 
-glucuronidase could
hydrolyze these conjugates, thereby liberating the genotoxic
N-hydroxylamine. However, whereas
N3-glucuronides can be deconjugated by
intestinal bacterial -glucuronidase, N2-glucuronides are resistant; thus,
generation of genotoxic N-hydroxy-PhIP by bacterial
-glucuronidase is unlikely to occur in the human intestine.
There is evidence from reconstitution and tissue culture studies that
N-hydroxy PhIP can also be sulfated (Buonarati et al., 1990;
Chou et al., 1995; Lewis et al., 1998). Three of the sulfotransferases (SULTs 1A2, 1A3, and 1E1) have been shown to sulfate heterocyclic amines and their hydroxylamine derivatives (Buonarati et al., 1990;
Ozawa et al., 1994; Lewis et al., 1998; Turesky et al., 1998). The
latter isoform, SULT1E1, is known to be hormonally regulated and
readily inducible by progesterone. This suggests that sulfation
activity could vary, for example during the luteal phase of the
menstrual cycle when there is a surge in progesterone levels and
SULT1E1 activity may be elevated (Lewis et al., 1998). Since the
sulfoxy ester of N-hydroxy-PhIP is an unstable product, its
detection in biological samples is likely to be very difficult. However, the 4'-PhIP-sulfate ester, a detoxication product, has been
detected in humans, thus demonstrating the involvement of sulfotransferase in PhIP metabolism (Malfatti et al., 1999).
N-Hydroxylation of the HAs, the primary metabolic pathway in
humans, is also the primary route of HA genotoxicity. For some HAs, the
N-hydroxy metabolite reacts poorly with DNA, but it can be
converted to highly reactive derivatives by esterification, e.g., to
form N-acetoxy, N-sulfonyloxy,
N-propyloxy, and N-phosphatyl derivatives (Schut
and Snyderwine, 1999). Studies with bacterial strains that are
deficient, proficient, and overexpress acetyltransferase enzymes show
the importance of this esterification reaction in the metabolic
activation of the heterocyclic amines to bacterial mutagens (see Fig.
4). Like the N-hydroxy sulfate
esters, the acetyl esters of the N-hydroxy heterocyclic
amines are extremely reactive and readily damage DNA.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 4.
Effect of acetyltransferase activity on the
bacterial mutagenicity of N-hydroxy PhIP.
Acetyltransferase activity is high in S. typhimurium
YG1024 strain, lower in TA98, and very low in TA98DNP.
|
|
| |
Genotoxicity |
Mutation.
In Ames S. typhimurium mutagenicity tests, the
HAs have been found to be extremely powerful promutagens causing both
frameshift and point mutation (Felton and Knize, 1990). The requirement
for metabolic activation and substantial species differences in this reaction affects the comparative genotoxicity of these compounds. Whereas the mouse is good at activating PhIP and MeIQx to genotoxic derivatives, other species such as the rat are poor (Davis et al.,
1993). However, activation of the HAs to genotoxic derivatives is
very efficient in humans due to high hepatic CYP1A2 activity (Edwards et al., 1994). Species differences in the ability to activate HAs are particularly pronounced between humans and cynomolgus monkey, since the liver of the latter is almost devoid of CYP1A2 expression and is consequently very poor at activating HAs to genotoxic
derivatives (Edwards et al., 1994).
Chemically induced mutation in bacterial cells is sometimes not
manifest in mammalian cell systems. Since the HAs are such powerful
bacterial mutagens, it was important to confirm their genotoxicity in
eukaryotic cells. A variety of Chinese hamster cells, both DNA
repair-proficient and -deficient have been used to study HA-induced
mutagenicity (Thompson et al., 1987). Interestingly, PhIP, a weaker
bacterial mutagen than MeIQx, is a particularly potent eukaryotic cell
mutagen (Thompson et al., 1987; Felton and Knize, 1990). To explore the
molecular nature of PhIP-induced mutation, we have used Chinese hamster
V79 cells and the hprt gene as a surrogate marker. Since V79
cells have no intrinsic CYP1A2 activity, culture with PhIP failed to
increase the mutation frequency above background spontaneous rate.
However a V79 cell line variant, engineered to express human CYP1A2
(XEMh1A2 cells) (Wölfel et al., 1992), showed a good PhIP
dose-mutation relationship (Yadollahi-Farsani et al., 1995). Upon
analysis, a high percentage of these mutants were G to T transversions
together with a significant number of 
1 G:C base pair deletions
(Table 3), in monotonous runs of G. The
preponderance of these guanine-based mutations occurred on the
nontranscribed strand (Yadollahi-Farsani et al., 1996).
The molecular nature of PhIP-induced mutation has also been examined in
an in vivo transgenic model, MutaMouse, in which 80 copies of the
lacZ bacterial transgene gene are present in every cell
(Lynch et al., 1996). The MutaMouse mice were administered oral PhIP
(20 mg/kg per day) for 4 days and then examined for mutation in the
intestinal tissue. Molecular analysis of the mutations confirmed a
remarkable similarity to those found in XEMh1A2 cells (see Table
3), viz., mainly G:C to T:A transversions with a significant number of 1 G:C base pair deletions (Lynch et al., 1997). Nearest neighbor analysis of PhIP-induced hprt mutants in XEMh1A2
cells and in the lacZ transgene showed that in both models
there was a preferred motif for mutation, a 5'-GGA-3' sequence. These
three characteristics of PhIP-induced mutation, i.e., G:C to T:A
transversion, 1 G:C frameshift mutation, and preference for 5'-GGA
motifs may constitute a mutational "signature" that is diagnostic
of the involvement of PhIP (Gooderham et al., 1997). Whether it is
possible to use mutational signatures to assess the involvement of
compounds such as PhIP in human disease remains to be established.
Carcinogenicity.
All the heterocyclic amines examined thus far have been shown to be
carcinogenic in bioassays (Ohgaki et al., 1991). Both MeIQx and PhIP
are carcinogenic in both mice and rats (Kato et al., 1988; Esumi et
al., 1989; Ito et al., 1991). In the mouse, MeIQx induces liver and
lung tumors, lymphomas, and leukemias, whereas in the rat it causes
tumors in the liver, the skin, the zymbal gland, and the clitoral gland
(Kato et al., 1988). In the mouse, PhIP induces predominantly lymphoma
(Esumi et al., 1989), but interestingly in the rat it induces tumors of
the colon and the prostate in the male and the colon and breast in the
female (Ito et al., 1991), the sites of the most common diet-related cancers found in humans in the Western world. Molecular analysis of the
colon tumors induced in male rats given PhIP showed that 5 of 8 had a
mutation in the Apc gene (Kakiuchi et al., 1995); in every
case the mutation was a 1 G:C base pair frameshift in a 5'-GGGA
sequence. Interestingly, mutation of Apc is a very common and early event in human colonic cancer (Powell et al., 1992). Thus,
the molecular nature of the mutations detected in these PhIP-induced
tumors was identical to the PhIP mutational signature in cultured cells
and transgenic mice.
| |
Conclusions |
There is now no doubt that humans who regularly consume cooked
meat are exposed to heterocyclic amines on a daily basis. These compounds are readily absorbed and bioavailable and are extensively, but variably, metabolized to derivatives that include the proximal genotoxic N-hydroxylamine compounds. Further metabolism of
the HAs generates conjugated metabolites that include glucuronides and
sulfate and acetyl esters, the esters being reactive derivatives capable of damaging DNA by forming adducts predominantly with guanine
bases. The consequence of this reaction with guanine is a
characteristic pattern of mutation. In the case of PhIP, the reproducibility of the nature and sequence context of these mutations are such that they form a mutational signature that is diagnostic of
its involvement. Such mutational signatures may be a means of
detecting the involvement of heterocyclic amines in mutating critical
gene targets such as cancer-associated genes.
The significance of these processes to the etiology of human cancer
requires further investigation. Within a population, exposure to the
heterocyclic amines will vary depending upon cooking practices and
preferences, the level of cooked meat consumption, and the degree of
affluence. The genotoxic potential of ingested HAs will depend upon the
balance between metabolic activation and detoxication, and the
involvement of polymorphic enzymes in some of these reactions would be
expected to influence susceptibility to the genotoxic outcome.
Subsequently, the site and extent of DNA damage incurred will depend
upon the efficiency of DNA repair mechanisms and the genetic
predisposition of the individual to accommodating or adapting to the
genetic damage. In line with these observations, in a study of
acetylation phenotype, meat intake, and risk for adenoma/colon cancer,
it was reported that fast acetylators have a higher odds ratio than
slow acetylators and that the odds ratio increases to >3 in fast
acetylators who have a high meat consumption (Roberts-Thomson et al.,
1996). In a separate study, Lang et al. (1994) reported that the
combination of rapid acetylation, rapid CYP1A2 activity, and
consumption of well done meat markedly increased the risk of colon
cancer (odds ratio, >6). Thus, the significance to humans of exposure
to the heterocyclic amines is likely to depend upon what combination of
these factors pertain to a particular individual, and by estimating the
relative contribution of each factor, we may be able to obtain better
estimates of overall risk.
These studies were supported by grants awarded by the Ministry
of Agriculture, Fisheries and Food, the Cancer Research Campaign, and
the World Cancer Research Fund.
Abbreviations used are:
HA, heterocyclic amines;
MeIQx, 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline;
DiMeIQx, 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline;
PhIP, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine;
SULT, sulfotransferase.
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
Abstract
Introduction
1 nmol
P450