Abstract
Background: 2,5-dimethylfuran (DMF) is formed during heating of foods. Following side chain hydroxylation, DMF could be a substrate for human sulphotransferases (SULTs), which may lead to formation of a DNA reactive electrophile. Only few conflicting in vitro and no in vivo studies on DMF currently exist. Materials and Methods: The tumorigenic potential of DMF was studied in multiple intestinal neoplasia ApcMin/+ (Min) mice that are sensitive to intestinal carcinogenesis and express hSULTs 1A1 and 1A2 (Min/hSULT). Min and Min/hSULT mice were orally exposed to DMF for six weeks. Results: The intestinal tumor development of untreated female Min/hSULT mice was significantly lower compared to that of untreated Min females. No such effects of hSULTs were seen in males. DMF had a weak tumorigenic potential in the colon of female Min/hSULT mice, but not in males. Tumor development in Min mice was not affected. Conclusion: Overall, the tumorigenic potential of DMF in a metabolically competent mouse model was not convincing.
Furan and its derivatives are so called food processing contaminants, which are formed in the Maillard reaction during food processing and can be found in a wide range of heat-treated food and beverages (1, 2). Furan, which is an established carcinogen, and several of its alkylated derivatives can be biotransformed into reactive metabolites (3). One such furan derivative is 2,5 dimethylfuran (DMF). DMF has been detected in heated foods such as roasted cocoa beans (4), canned foods (5), and both instant and ground coffee (5, 6), and it has previously been used as a food flavouring agent. The Joint FAO/WHO Expert Committee on Food Additives and the European Food Safety Authority have expressed a safety concern due to sparse data on in vivo genotoxicity of DMF, which precluded its evaluation and approval (7, 8). Further studies on toxicity and genotoxicity were required.
Mostly in vitro experiments have been published regarding the genotoxicity of DMF. DMF did not induce reverse mutations in the Ames test (9-11), while in the bacterial “rec assay” there were some positive findings (11). Studies with mammalian cells provided inconclusive results on the induction of chromosomal aberrations after DMF exposure (12, 13). Finally, a recent publication by Fromowitz et al. (14) suggested clastogenic and genotoxic effects of DMF on isolated bone marrow cells from C57BL/6J (B6) mice. In this in vitro micronucleus test, increased formation of micronuclei was detected in response to increasing concentrations of DMF (0.1 and 0.5 mM). Recently, the reactivity of DMF were confirmed in vivo, where protein adducts were detected after intraperitoneally treatment of mice with DMF (15).
There exists a lack of information regarding absorption and metabolic fate of DMF in vivo. Studies on structurally-related furan derivatives, such as furfural, furfuryl alcohol, 2-methylfuran and 5-hydroxymethylfurfural, show that this group of substances is readily absorbed from the gastrointestinal tract of rodents and eliminated via urine (16-20). It is reasonable to assume that DMF is absorbed and eliminated in a similar fashion. As for the other furan derivatives, DMF bioactivation might occur in two steps: hydroxylation followed by sulphonation of one of the exocyclic methyl groups (Figure 1). Sulpho conjugation catalyzed by sulphotransferases (SULTs) is normally a detoxification pathway, but spontaneous cleavage of the sulpho group leaves a reactive electrophile that can react with proteins or DNA (21-23). Such bioactivation has recently been demonstrated in vitro for the structurally-related compounds 5-hydroxymethylfurfural and furfuryl alcohol (24, 25). The hSULT1A1 enzyme has broad substrate specificity (26, 27) and may be a potential catalyst for this bioactivation. For DMF, cytochrome P450 (CYP) mediated hydroxylation of a methyl side chain would be a prerequisite for sulpho conjugation (7).
A hypothetical pathway for bioactivation of 2,5-dimethylfuran (DMF). Cytochrome P450 (CYP) may oxidise a methyl side chain creating 5-methyl-2-furanmethanol, that could be a substrate for sulphotransferases (SULTs) in the presence of co-factor 3’ phosphoadenosine 5’ phosphosulfate (PAPS). Spontaneous cleavage of the sulpho conjugate leaves a reactive intermediate that may exert genotoxic effects. The pathway is deduced from information on similar substances as reviewed by the European Food Safety Authority (43).
The aim of the present study was to examine the tumorigenic potential of DMF. We used the ApcMin/+ (Min) mouse, that is a model for the human familial adenomatous polyposis cancer syndrome (28, 29). The Min mouse spontaneously develops numerous intestinal tumors, and is highly susceptible to intestinal carcinogens (28). This model develops colonic flat aberrant crypt foci (ACF) in addition to classical ACF. Flat ACF are dysplastic tumor precursor lesions (30). As there is a considerable species difference in the substrate specificity and tissue distribution of many SULT forms, standard rodent models may not be appropriate to predict the human risk of DMF. For this reason, the Min mouse was “humanized” by introducing the human SULT1A1-1A2 gene cluster to increase its capacity, i.e. in the intestine, to bioactivate hydroxyl methyl substituted furans.
Materials and Methods
Chemicals. DMF (CAS no. 625-86-5, 99% purity) was purchased from Sigma Aldrich (St. Louis, MO, USA) and diluted in corn oil intended for human consumption (Eldorado, NorgesGruppen, Oslo, Norway) shortly before use. For the genotyping, SDS and Proteinase K were purchased from Sigma Aldrich, GoTaq polymerase and all dNTPs from Promega Biotech AB (Nacka, Sweden), PCR buffer II and MgCl2 from Applied Biosystems (Foster City, CA, USA), and primers from Eurogentec (Seraing, Belgium). The FlashGel system was obtained from Lonza (Basel, Switzerland).
Animals. C57BL/6J (B6) multiple intestinal neoplasia ApcMin/+ (Min) mice, heterozygous for a mutation in the tumor suppressor gene adenomatous polyposis coli (Apc), were bred at our institute from B6 Min males and B6 wild type (wt) females, both originally purchased from the Jackson Laboratories (Bar Harbour, ME, USA).
B6 mice expressing human sulphotransferases 1A1 and 1A2 (hSULT mice) were generated at the Norwegian Institute of Public Health by backcrossing transgenic FVB/N (FVB) hSULT mice constructed at the German Institute of Human Nutrition (31) onto the B6 background. In short, hemizygous male FVB hSULT mice descending from a line containing multiple copies of the human SULT1A1-1A2 gene cluster, termed tg1 in the original publication (31), were crossed with B6 wt females. Male pups carrying the transgene were subsequently mated with B6 wt females for at least eight generations to ensure successful backcrossing of the hSULT genotype on the B6 background.
Breeding trios of two female hemizygous hSULT mice and one male Min mouse were used for breeding of experimental animals. This breeding gave rise to offspring of four genotype combinations: Apc+/+ without hSULT (wt), ApcMin/+ without hSULT (Min), Apc+/+ with hSULT (hSULT), and finally ApcMin/+ with hSULT (Min/hSULT) (Figure 2).
At two to three weeks of age the pups were earmarked for identification, and ear cartilage was used for genotyping. Only Min and Min/hSULT pups were included in the experiment. Litters were separated from their mother at three weeks of age, and co housed with littermates of the same sex, one to five animals in each cage. All mice were fed a standard pellet rodent diet (HT2019, Harlan Teklad Industries Inc., Indianapolis, IN, USA) and given tap water, both ad libitum. There was a 12-h light/dark cycle and the mice were housed in plastic cages on Nestpak Aspen 4HK bedding (Datesand Ltd., Manchester, UK). The experiments were carried out in conformity with the laws and regulations for experiments with live animals in Norway, and were approved by the Norwegian Animal Research Authority.
Overview of the four genotypes generated in the experiment and the groups included in the experiment. A heterozygous ApcMin/+ male was mated with hemizygous human SULT1A1/1A2 (hSULT) females generating wt, Min, hSULT and Min/hSULT pups.
Genotyping. The genotypes of the pups were identified with an allele-specific polymerase chain reaction (PCR) assay on DNA isolated from ear cartilage. The biopsies were each submerged in 60 μl TE-buffer (10 mM Tris pH 7.4, 0.1 mM EDTA, pH 8) with 0.05% SDS and incubated at 95°C for 10 min. A final concentration of 0.91 mg/ml Proteinase K was added before incubation at 55°C overnight. A final incubation at 95°C for 10 min was conducted to inactivate proteinase K and samples were stored at −20°C until PCR amplification. The protocol for detection of the Min genotype has been described previously (32). To determine the hSULT genotype, the PCR reaction was conducted in a 10 μl reaction volume containing 5 μl of a 1:100 dilution of the isolated DNA, 0.2 μM of each primer h1A1Ex5F (5’-TCCTGGAGAAGTTCATGGTC -3’) and h1A1Ex6R (5’ CACCACTCCTGCACGTGCTG-3’), 0.2 mM each of dCTP, dGTP, dTTP and dATP, 1 × PCR buffer II (10 mM Tris-HCl, pH 8.3, 50 mM KCl), 1.5 mM MgCl2 and 0.025 U GoTaq polymerase. PCR was conducted in an iCycler Thermal cycler (Bio-Rad, Hercules, CA, USA) as follows: 2 min at 94°C (hot start) and 30 cycles of 94, 58 and 72°C, each for 30 sec, followed by a final extension at 72°C for 7 min. The PCR product was visualised by electrophoresis in 2.2% agarose gels and identified as a 162-base pair bond.
Treatment. Min and Min/hSULT pups were gavaged with 5 mg DMF/kg body weight (bw) (LowDMF), 25 mg DMF/kg bw (MedDMF) or 50 mg DMF/kg bw (HighDMF), or vehicle (Ctrl) three times a week from week three until week eight. Each animal received a total of 18 administrations, and bw was recorded each time. The mice were euthanised by cervical dislocation at ten weeks of age and terminal bw was recorded along with the weight of the spleen and kidney.
Scoring of tumors and flat aberrant crypt foci (ACF). The small and large intestine were removed separately, rinsed in ice-cold PBS and incised longitudinally. The small intestine was divided into proximal, middle and distal parts. The pieces were then spread flat between wet filter papers and fixed in 4% neutral-buffered formalin solution for at least 48 h. The intestines were stained for approximately 5 sec in 0.2% methylene blue (George T. Gurr Ltd., London, UK) and stored in 4% neutral buffered formalin until the mucosa was examined by transillumination using an inverse light microscope. The number, size and location of lesions were recorded. Tumor size was scored as mm2, and the location of the lesions in the small intestine and colon were referred to as cm distal from the ventricle and caecum, respectively. The tumor load (tumor size × number of tumors) was calculated for both the small intestine and colon. Scoring of flat ACF in the colon was performed as previously described (33), and the size of an ACF was scored as the number of crypts. Lesions with >10 aberrant crypts were defined as tumors.
Statistical analysis. Litters were randomly assigned to treatment groups. A mean value of measurements from littermates of the same sex and genotype (1-4 pups) was used as the experimental unit. Differences between the groups were evaluated with a two-way analysis of variance test, applying the Holm-Šídák method for comparison against control, in SigmaPlot version 12.0 (Systat Software GmbH, Erkrath, Germany). When the Shapiro-Wilk test indicated non-normally distributed data, a manual ranking of the dataset was performed prior to two-way analysis of variance. Data from individual animals, not litters, were used for calculation of incidence of tumors or flat ACF in the colon (the proportion of individuals with one or more lesions). Differences in incidence between the groups were analysed with the z-test. A p-value ≤0.05 was considered significant.
Results
Effect of the hSULT genotype on spontaneous tumor development in Min mice. The tumor development in untreated Min/hSULT females was significantly lower compared to untreated Min females. Small intestinal tumors were significantly smaller, but not fewer, in Min/hSULT females in comparison to Min females (0.48 mm2 vs. 0.65 mm2, p=0.002) (Table I). In addition, both the incidence of colonic tumors and the colonic tumor load in Min/hSULT females were significantly lower than those in Min females (38% vs. 82%, p=0.037, and 0.0 mm2 vs. 5.8 mm2, p=0.001, respectively) (Figure 3 and Table I). For the untreated males, there were no differences in the tumor development between Min mice and the Min/hSULT mice (Table II).
Tumours and flat aberrant crypt foci (ACF) in the small intestine and colon of female ApcMin/+ mice with (Min/hSULT) and without (Min) human SULT1A1/1A2 exposed to 5 (LowDMF), 25 (MedDMF) or 50 (HighDMF) mg DMF/kg body weight (bw) or vehicle (Ctrl).
Tumorigenicity of DMF in Min mice. No effect of DMF treatment was detected in female Min mice (Table I). A possible minor effect of DMF exposure on tumorigenicity in Min males was observed in the form of a tendency of increased incidence of colonic tumors in exposed mice (p=0.087) (Table II) (Figure 3B).
Tumorigenicity of DMF in Min mice with expression of hSULT1A1/1A2. In the Min/hSULT females there was a small, but statistically significant, increase in tumor development after DMF exposure. There was an increase in incidence of colonic tumors in Min/hSULT females, from 38% in the untreated group to 84% in the MedDMF group (p=0.022) (Figure 3A). A similar, but non-significant increase in incidence was seen for the two remaining DMF-exposed groups when compared to untreated Min/hSULT females (Figure 3A). The colonic tumor load showed the same tendency, but also in this case the medium-dose group (6.1 mm2) was the only group significantly different from untreated females (0.0 mm2, p=0.015) (Table I). Figure 4 depicts the distribution of tumors along the length of the small intestine and colon of female Min/hSULT mice. The figure shows that after DMF exposure the tumor number increases in the middle to distal part of the small intestine and colon, in the same area where most of the spontaneous tumors are found. These effects of DMF in Min/hSULT females could alternatively be due to the low background level of tumors in the untreated females, which was described in section 3.1. There was no effect of DMF on the tumor development in Min/hSULT males (Table II) (Figure 3B), nor on flat ACF development in any of the sexes (Table I and II).
Body weight gain and relative spleen and kidney weight. Body weight (bw) was recorded each week for all mice, and no significant differences were observed between the groups at any age (data not shown). The percentage of bw gain at the end of the experiment was similar between exposed and unexposed Min mice, and also between exposed and unexposed Min/hSULT males (Table III). However, Min/hSULT females exposed to the LowDMF (84%) and HighDMF (78%) doses had a significantly lower final percentage bw gain than the untreated Min/hSULT females (10.2%, p=0.038 and p=0.024, respectively) (Table III). There was also a tendency for the MedDMF treatment group to have a reduced bw gain (85%) compared to the untreated group (p=0.074) (Table III).
Tumours and flat aberrant crypt foci (ACF) in the small intestine and colon of male ApcMin/+ mice with (Min/hSULT) and without (Min) human SULT1A1/1A2 exposed to 5 (LowDMF), 25 (MedDMF) or 50 (HighDMF) mg DMF/kg body weight (bw) or vehicle (Ctrl).
The percentage relative spleen and kidney weight was calculated. There was no difference in relative spleen weight following DMF exposure for any of the Min mice, or between exposed and untreated male Min/hSULT mice (Table III). However, relative spleen weight in female Min/hSULT mice was increased in exposed mice. The difference was significant for mice exposed to LowDMF (0.65%, p=0.026), MedDMF (0.65%, p=0.013) and HighDMF (0.62%, p=0.037) compared with untreated mice (0.45%), but there was no dose-response. The relative weight of the kidney did not differ between the treatment groups (data not shown).
Discussion
In the present study we examined the tumorigenicity of oral exposure to the food processing contaminant and food flavouring DMF in Min mice expressing hSULT1A1/1A2. Expression of hSULT1A1/1A2 in untreated Min females reduced the size of the small intestinal tumors, as well as the incidence of colonic tumors and colonic tumor load, in comparison to untreated regular Min females. No such effects were seen in males. There was a modest increase in tumor development of DMF-exposed Min/hSULT females, including an increased incidence of colonic tumors and colonic tumor load.
The hSULT mouse model on the B6 genetic background was generated for this experiment and no description of this strain has been previously published. Additionally, the original hSULT mouse model on the FVB background was made only a few years ago (31) and has rarely been used for tumorigenicity studies (34). Only once before has the FVB hSULT mouse been crossed with the Min mice (35). The present experiment is also the first report on tumorigenicity of DMF in mice.
DMF did not seem to increase the tumor development in the Min mouse model. This is in accordance with results from our laboratory indicating no genotoxicity of DMF in the in vivo alkaline single-cell gel electrophoresis assay after oral exposure (36), even though genotoxicity has been reported in vitro (14). A dose-dependent increase in DMF protein adducts were recently detected in mice, supporting the reactivity of DMF. However the DMF were administered intraperitoneally and therefore cannot be directly compared to the results from the Comet assay (15). Looking at the tumor incidence in untreated Min females, 14 out of 17 mice (82%) developed colonic tumors. This is a high background level that limits the possibility to detect further increase in tumor incidence by DMF.
Percentage body weight (bw) gain and relative spleen weight in ApcMin/+ mice with (Min/hSULT) and without (Min) human SULT1A1/1A2 exposed to 5 (LowDMF), 25 (MedDMF) or 50 (HighDMF) mg 2,5-dimethylfuran (DMF)/kg bw or vehicle (Ctrl).
There was a statistically significant effect of DMF on the tumor development in the colon of Min/hSULT females. The incidence of colonic tumors and the colonic tumor load of exposed Min/hSULT females were increased to the levels seen in Min females. The observed impact of exposure in Min/hSULT females seems to be detected mainly because of the low background level in the untreated females of the same genotype. The reduced weight gain and increased relative spleen weight may suggest an increased sensitivity toward DMF for the Min/hSULT females, although the effect was not clearly dose-dependent. Since no mechanistic studies with DMF and SULTs exist, there is no supporting evidence for DMF bioactivation by these enzymes. In our study we cannot exclude the possibility that the observed effects were attributed to other metabolites of DMF than those hypothetically formed by side chain oxidation and subsequent sulphonation by hSULT1A1/1A2.
The significantly lowered tumor development in untreated Min/hSULT females compared to untreated Min females was an interesting finding. Apparently the hSULT genotype offers some kind of protection against tumor development, perhaps by increased inactivation of endogenous or environmental compounds. Untreated male Min/hSULT mice did not show a similar effect on tumor development. There is a sex-related difference in the expression of several Sult enzymes in rodents. For most sex-divergent Sults there is a female predominance (37), and the difference in murine Sult mRNA levels are probably caused by suppressive effects of both androgens and male-pattern growth hormone secretion (38). However, the FVB hSULT mice show minimal variance between males and females in hSULT1A1/1A2 protein levels (31). The sulphonation activity in cytosolic preparations, however, did not always reflect the level of enzyme. In addition, the influence of sex-specific hormones like oestrogen in tumorigenesis should be considered, since hSULT1A1 catalyze the sulphonation of oestrogens to form water-soluble and biologically inactive oestrogen sulphates (26). High levels of oestrogens are associated with female cancers such as breast, ovarian and endometrium (39), but have been implicated in reduction of tumor formation in Min mice (40).
Incidence of colonic tumors in (A) female and (B) male ApcMin/+ mice with (Min/hSULT) or without (Min) human SULT1A1/1A2 exposed to 5 (LowDMF), 25 (MedDMF) or 50 (HighDMF) mg DMF/kg bw or vehicle (Ctrl).
Smooth curves of the distribution of tumors (mean) along the length of the (A) small intestine and (B) colon of female ApcMin/+ mice with human SULT1A1/1A2 (Min/hSULT) mice exposed to 5 (LowDMF), 25 (MedDMF) or 50 (HighDMF) mg DMF/kg bw or vehicle (Ctrl). Location of tumors in the small intestine is given as cm distal to the ventricle, and colonic tumours as cm from the caecum.
In previous studies in our laboratory the tumor development in Min/hSULT mice on the FVB background did not differ between sexes (35). A direct comparison between these strains, however, is difficult, as FVB Min mice spontaneously develop considerably less tumors than B6 Min mice (41). We also note that another study on hSULT-expressing mice reported higher levels of DNA damage in females compared to males following exposure to furfuryl alcohol (42). That study did not provide any explanation on this finding, but supports the hypothesis that there are aspects of SULT activity that differ between females and males.
The high tumor development in untreated animals with a tumor incidence of up to 80-90% could prevent the detection of tumor development from DMF exposure, if the tumor development in the mouse model approaches a maximum level. DMF should, therefore, also be studied in mice models with lower background development of tumors, before a final conclusion on the tumorigenic potential of DMF can be drawn.
In conclusion, untreated female Min/hSULT mice seemed to be influenced by the hSULT genotype in the form of reduced spontaneous tumorigenicity. Possibly, this low tumorigenicity in untreated mice might have led to detection of a statistically significant tumor development in the Min/hSULT females following DMF exposure, although an effect of DMF cannot be fully excluded. The present study, however, provides no convincing evidence of DMF as a potent carcinogen, neither in regular Min mice nor in Min mice expressing human SULT1A1/1A2.
Acknowledgements
The authors wish to thank Victor Labay Ong, Hege Hjertholm and Trude Karin Olsen at the Norwegian Institute of Public Health for their excellent technical assistance. The hSULT mouse on the FVB background used to generate the hSULT mouse on the B6 background was a generous gift from Prof. Glatt at the German Institute of Human Nutrition.
Footnotes
This article is freely accessible online.
- Received December 11, 2015.
- Revision received January 14, 2016.
- Accepted January 15, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.