Prenatal exposure of mice to the human liver carcinogen aflatoxin B1 reveals a critical window of susceptibility to genetic change

It has become axiomatic that critical windows of susceptibility to genotoxins exist and that genetic damage in utero may be a trigger for later life cancers. Data supporting this critical window hypothesis are remarkably few. This study provides a quantitative bridge between DNA damage by the liver carcinogen aflatoxin B1 (AFB1) during prenatal development and the risk of later life genetic disease. AFB1 was given to pregnant C57BL/6J mice, carrying F1 gestation day 14 (GD14) embryos of the B6C3F1 genotype. Ultra‐high performance liquid chromatography and mass spectrometry (UPLC‐MS) using aflatoxin‐15N5‐guanine adduct standards afforded measurement of the AFB1‐N7‐Gua and AFB1‐FAPY adducts 6‐hr post dosing in liver DNA of mothers and embryos. A parallel cohort gave birth and the livers of the F1 were analyzed for mutations in the gpt gene at 3 and 10 weeks of age. The data revealed mutational spectra dominated by G:C to T:A mutations in both the mother and offspring that are characteristic of AFB1 and distinct from background. It was shown that adducts in GD14 embryos were 20‐fold more potent inducers of mutagenesis than adducts in parallel‐dosed adults. This sensitivity enhancement correlated with Ki67 staining of the liver, reflecting the proliferative potential of the tissue. Taken together, these data provide insight into the relative genetic risks of prenatal and adult exposures to AFB1. Early life exposure, especially during the embryonic period, is strikingly more mutagenic than treatment later in life. Moreover the data provide a baseline against which risk prevention strategies can be evaluated.

Factors including rapid cell division and unique developmental changes during the prenatal period create a remarkable window of susceptibility to the numerous toxic effects of environmental agents. Accordingly, maternal exposures to environmental factors during pregnancy influence the risk of many chronic adult-onset diseases in the offspring, including cancer. [1][2][3] Early life including in utero exposures can also result in growth impairment, behavioral defects and teratogenesis. 4,5 AFB 1 is a toxin produced by fungi that contaminate food supplies especially in the developing world. It is a potent genotoxin that induces hepatocellular carcinoma (HCC) in most animal species, including humans. 6 In sub-Saharan Africa and many parts of Asia where 80% of HCC cases occur, there is growing evidence that AFB 1 exposure disproportionately affects childhood health in these regions. 7 The toxic properties of AFB 1 require metabolic activation in the liver to the reactive exo-8,9-epoxide by cytochrome P450s.
The epoxide reacts at the N 7 atom of guanine in DNA to form an initial AFB 1 -N 7 -Gua adduct, [8][9][10][11][12] which can later be converted by hydrolysis of the imidazole ring to the chemically and biologically more stable AFB 1 -FAPY adducts, or by hydrolysis of the glycosidic bond to yield an abasic site. 6 One of the FAPY adducts is thought to be the principal driver of AFB 1 mutations, 13 which is primarily the G to T transversion. G:C to T:A is also the dominant mutation type observed in the liver tumors of people exposed to AFB 1 . 14 A biochemical strategy that blocks AFB 1 -DNA damage involves AFB 1 -glutathione conjugation by alpha class glutathione-S-transferases and other Phase II enzymes that effectively shunt the AFB 1 -8,9-epoxide from DNA, 15,16 rendering the toxin harmless in vivo. While mice start life very sensitive to AFB 1 , they become increasingly refractory over the first 2 weeks of life as the levels rise of the enzymes that destroy the epoxide, 17 and this biochemical reprogramming is likely to be one factor that renders the adult mouse completely insensitive to carcinogenesis by AFB 1 when treated during late adolescence or adulthood. 17 Studies by Vesselinovitch and Wogan showed that a single exposure of the B6C3F1 mouse to AFB 1 during the neonatal period, days 1-7 after birth, results in a high incidence of HCC in adulthood. 18 By contrast adult mice, post day 28, exposed to the same dose do not develop HCC at any age. The high initial sensitivity to AFB 1 that tapers off over the first 2 weeks of life makes the B6C3F1 mouse a good model to understand how specific host variables, such as toxin metabolism and DNA adduct formation, relate to later life biological outcomes, including growth stunting, cognitive defects and genetic diseases including cancer. This model also provides an opportunity to probe the basis for the selective sensitivity of males to carcinogenesis. Newborn B6C3F1 mice treated with single or multiple doses of AFB 1 develop HCC by 82 weeks of age in >90% of the males, compared to <10% of identically treated females. 18 The high sensitivity of newborn mice to AFB 1 prompts the question as to whether and to what degree AFB 1 also poses a risk to the fetus during gestation. In utero risk is suggested by the detection of AFB 1 -DNA adducts in human cord blood and placental tissue. 19,20 AFB 1 has been shown to act as a transplacental carcinogen in Sprague Dawley (SD) and Wistar rats 21,22 and in hepatitis B transgenic mice. 23 Epidemiological and animal studies have established AFB 1 -DNA adducts as valuable predictors of carcinogenic risk in experimental animals and in humans. 24 Adducts are widely used biomarkers because they are believed to be the chemical precursors to the genetic changes that initiate or facilitate the conversion of normal cells to cancer cells. 14 To date there have been no studies quantitatively linking DNA adducts and mutagenic endpoints in mice treated in utero. Recent work has shown that the AFB 1 -N 7 -Gua and AFB 1 -FAPY adducts are formed in the liver DNA of B6C3F1 mice treated with AFB 1 shortly after birth, at days 1-4 of life. Adduct levels in the neonates correlate with the frequency, types and distributions mutations in liver tissues of 3-week and 10-week-old mice. 25 The predominant mutation in these animals is the G:C to T:A transversion, which is a signature of genetic damage produced by aflatoxin exposure. 25 To quantify the risk of even earlier life exposure to AFB 1 , in the present study we measured AFB 1 -DNA adduct levels, and characterized the subsequent frequency and spectrum of mutations arising in fetus-exposed livers at 3 and 10 weeks of age. Although several studies have examined mutations in mice treated with chemicals in utero, the quantitative relationships between chemical damage to DNA during fetal development and future health risks are largely unexplored. B6C3F1 mice were used, because they are known to be sensitive to AFB 1 -induced HCC and, of great importance, they are widely used in regulatory analyses as the principal murine model for assessment of chemicals for carcinogenic activity. The strain of B6C3F1 mice used was molecularly equipped with a gpt transgene that can use as reporter gene to detect base pair substitutions and deletions. The present study builds on earlier work relating DNA adducts and human HCC susceptibility 26 ; it reveals, using an animal model, a strikingly strong early-life vulnerability to a naturally occurring carcinogen, using DNA adducts and gpt mutations as predictive metrics of later-life susceptibility.

Material and Methods
Chemicals AFB 1 was purchased from Sigma Chemical (St. Louis, MO). Unless otherwise specified, all reagents were at least of ACS reagent grade.

Animals
C57BL/6J gpt delta transgenic mice were obtained from Dr. Takehiko Nohmi. 27 The gpt delta B6C3F1 mice used in our experiment were generated by breeding female gpt delta What's new? Genetic damage in utero can pave the way for cancer later in life. Why are younger animals more susceptible to carcinogens than adults? In this paper, the authors quantify this effect for the first time. They exposed fetal mice to aflatoxin and compared the rates of cancer-causing mutations with adult mice exposed to the same dose of aflatoxin. The DNA adducts created were 20-fold more likely to cause mutations in liver cells in the young mice than in the adults. Not surprisingly, they observed that fetal liver cells were dividing much faster than adult liver cells, providing evidence for the view that rapid cell division creates this youthful susceptibility to carcinogens.
C57BL/6J mice, which harbor an estimated 80 copies of the gpt gene on chromosome 17, with male C3H/HeJ mice that were purchased from the Jackson Laboratories (Bar Harbor, Maine). Pregnant gpt delta C57BL/6J mice were administered AFB 1 (6 mg kg 21 dissolved in 100 lL of DMSO for mutation analysis and 5 mg kg 21 for adduct analysis) via intraperitoneal (i.p.) injection or oral gavage on gestation day 14 (GD14); before being dosed with AFB 1 , the mothers used in this study had all given birth to at least one litter prior to inclusion in the study. Control animals received 100 lL DMSO only. At the indicated times ( Fig. 1), mice were euthanized and liver tissue was collected for isolation of DNA. All experiments were conducted in accordance with protocols approved by the MIT Committee on Animal Care.

DNA adduct analysis in liver tissues
DNA for AFB 1 adduct analysis was isolated from nuclei using previously described methods. 28 AFB 1 -DNA adducts were hydrolyzed by treatment with 0.1 N HCl at 95 C for 15 min. 9 After hydrolysis to liberate AFB 1 -DNA adducts, internal 15 N 5guanine-derived standards were added to permit quantitative analysis by isotope dilution mass spectrometry for both AFB 1 -N 7 -Gua and AFB 1 -FAPY. Adducts were separated by UPLC-MS. The protonated parent ion of the AFB 1 -N 7 -Gua adduct (m/z 480.1) was selected and subjected to collision-induced fragmentation producing a m/z 152 product ion that was monitored to quantify adduct levels. The AFB 1 -FAPY adduct was measured by selection of the m/z 498 parent ion and monitoring the collision-induced product ion m/z 452. 29

DNA isolation and in vitro packaging
Liver tissue was pulverized in a mortar and pestle under liquid nitrogen and stored at 280 C. Genomic DNA was extracted from 25 mg of tissue using the RecoverEase DNA Isolation Kit (Agilent Technologies, Santa Clara, CA). k-EG10 phages were packaged in vitro from genomic DNA using Transpack Packaging Extract (Agilent Technologies) following the manufacturer's instructions.

Gpt mutation assay and sequencing analysis
Mutations in the gpt gene were identified by selection with 6thioguanine (6-TG), as previously described. 27 Briefly, k-EG10 phages rescued from genomic DNA were transfected into E. coli YG6020 expressing Cre recombinase and cultured on selective media containing chloramphenicol (Cm) and 6-TG or Cm alone for 72 hr until the appearance of colonies. 6-TG resistance was confirmed by regrowth of colonies on plates containing Cm and 6-TG. For sequencing, plasmid DNA was isolated using a Qiagen Miniprep Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Sequencing of the gpt gene was performed by the Biopolymers Facility at Harvard Medical School (Boston, MA) on a 3730xL DNA Analyzer (Applied Biosystems, Foster City, CA) using the forward primer: 5 0 TCTCGCGCAACC-TATTTTCCC -3 0 . Sequencing data were aligned with the E. coli gpt gene (GenBank M13422.1) using NCBI Nucleotide Blast. Mutation frequencies (MFs) were calculated as the number of confirmed 6-TG mutant colonies divided by the total number of Cm-resistant colonies.

Immunohistochemical staining of mouse liver for Ki67
Liver specimens were fixed in formalin, embedded in paraffin and cut into 4 lm sections. Following deparaffinization, rehydration, and antigen retrieval with citrate buffer, endogenous peroxidase activity was blocked (DAKO 3950, Agilent Technologies). Sections were incubated for 60 min with a Ki67 antibody (Santa Cruz Biotechnology, USA; sc-15402) followed by a secondary biotinylated antibody (Rabbit anti-mouse IgA-B, Santa Cruz Biotechnology, USA; sc-358961). Immune complexes were visualized by addition of streptavidin conjugated horseradish peroxidase and diaminobenzidine (DAB) reagent. Tissues were counterstained with hematoxylin. A proliferation index was determined as the number of Ki67 positive cells in 10 high power fields (403) divided by the total number of observed cells.

Statistical analysis
Student two-tailed t tests were used to analyze DNA adduct and mutation frequency data. Analyses were done with GraphPad Prism software and a value of less p < 0.05 was considered statistically significant. 26,30 The Mann-Whitney nonparametric test was used to determine the statistical significance of differences in MF.

Results
Mutagenic AFB 1 -DNA adducts form in both fetal and maternal liver DNA adducts formed by AFB 1 are potent inducers of G:C to T:A mutations, 14,31 which is among the most frequent mutations found in human HCC. 14 To determine whether the developing fetus is at risk of genetic damage from maternal exposures to AFB 1 , pregnant gpt delta C57BL/6J mice at gestation day 14 were treated with the toxin and the levels of AFB 1 -DNA adducts in both maternal and B6C3F1 fetal livers were quantified 6 hr after administration of AFB 1 via i.p. injection, providing a snapshot of adduct density at a time when it is reasonable to assume that carcinogenesis is initiated. The experimental scheme is shown in Figure 1. The amounts of AFB 1 -N 7 -Gua and AFB 1 -FAPY adducts were analyzed by quantitative isotope dilution mass spectrometry (Table 1). 29 The highest levels of both adduct types were found in the C57BL/6J maternal DNA where there was approximately two-fold greater amounts of the persistent AFB 1 -FAPY compared to AFB 1 -N 7 -Gua. These observations are consistent with the rapid removal of the latter adduct as well as its conversion to the more chemically stable AFB 1 -FAPY by chemical hydrolysis of the imidazole ring. 28 Lower but measurable levels of AFB 1 adducts were found in B6C3F1 fetal liver DNAs. A total of 12 fetal livers; six from each of two litters were analyzed. Within each litter there was a good correlation of adduct levels among individuals with equal amounts of the AFB 1 -N 7 -Gua and AFB 1 -FAPY adducts. Overall, fetal liver adduct levels (0.62 adducts/10 6 nucleotides) were 100-fold lower than in maternal liver (63.5 adducts/10 6 nucleotides). These results indicate that the toxin reaches the liver, and that the developing liver in utero is capable of metabolizing AFB 1 to its reactive exo-8,9-epoxide, which has the ability to form covalent DNA adducts with potentially mutagenic properties. 13,31 One potential complication of the experiment described above is that compounds administered to pregnant animals via i.p. injection could directly access the uterine compartment without entering the maternal circulation. To understand how this possibility might influence adduct levels, in a separate experiment, pregnant C57BL/6J mice were treated with AFB 1 by gavage and once again DNA adducts were measured after 6 hr. After oral administration we found that the AFB 1 -N 7 -Gua and AFB 1 -FAPY adducts were present in maternal liver in the same proportion as found after i.p. injection; however, total adduct levels by this route, were about 2.5-fold lower than levels after i.p. administration. Both of the AFB 1 -DNA adduct types were also observed in the fetal DNA liver samples of gavage-treated animals, but only the AFB 1 -N 7 -Gua was present at amounts above the limit at which confident quantitation was possible. The levels of this adduct were about four-fold lower than those found in fetal DNA after i.p. dosing. Differences in the rates of absorption, distribution and metabolism of AFB 1 are likely responsible for the lower levels of AFB 1 -DNA adducts following gavage administration. Based on this experiment, in which AFB 1 -DNA adducts were observed in fetal liver after treatment via either i.p. injection or by oral administration it is clear that the developing liver is at risk for genetic changes or toxicity that could predispose animals to develop later life diseases.
Aflatoxin is mutagenic to GD14-embryonic mice, as detected 10 weeks after birth The experiment above showed that DNA adducts form in the fetus, but at only about 1% the absolute level that occurs in the mother. We next probed whether the low but finite level of adduct formed in the fetus posed a significant threat for mutations that could be detected in adolescence or adulthood (at 3 or 10 weeks of age). The gpt transgenic locus in the gpt delta C57BL/6J mouse was used to determine the impact of AFB 1 -DNA damage in utero on the mutant frequency in the livers of hybrid B6C3F1 mice. Pregnant gpt delta C56BL/6J mice were treated on GD14 with 6 mg kg 21 AFB 1 via i.p. injection; control animals received the DMSO vehicle only. The E. coli gpt reporter gene was rescued from genomic DNA, and 6-TG resistant mutants were enumerated. This assay is sensitive to point mutations, such as base substitutions, and frameshifts resulting from small deletions and insertions, which are characteristic of the AFB 1 -induced mutational spectrum. 14,31 Mothers gave birth to B6C3F1 pups at around GD21. There were no observable phenotypic differences between AFB 1 -treated and control litters. In an initial experiment, mothers and pups were sacrificed at weaning, 3 weeks of age. Genomic DNA was isolated from liver tissue and mutational analysis was performed on the gpt transgene. At 3 weeks of age (Fig. 2a), we did not observe a statistically significant increase of AFB 1 -induced mutations in the B6C3F1 above the baseline level in control (DMSO-treated) animals (both males and females).
At 10 weeks of age (Fig. 2b), however, both males and females treated at GD14 showed a significantly elevated MF in AFB 1 -treated over control animals. We found that exposure to AFB 1 on GD14 statistically increased the gpt MF by threefold in 10-week-old male mice (p 5 0.001) and 2.5-fold in 10-week-old females (p 5 0.003) compared with DMSOtreated controls. Overall, males and females, when treated in utero, appear to be equally sensitive to the mutagenic properties of AFB 1 at the 10-week time point. Figure 3 examines the gpt MFs of the C57BL/6J mothers of the B6C3F1 mice described above. The mothers followed the same pattern as their offspring, with significant enhancement of AFB 1 -induced mutations seen at 10 weeks after parturition (p 5 0.016). As indicated above, the mothers had 1003 higher levels of DNA damage than the GD14 fetus, yet comparison of Figures 2b and 3 reveals only a 4.6 fold difference in the absolute MF between the mothers and fetuses (based upon MF measurements at 10 weeks post birth). The reasons for the apparently enhanced mutagenic potency of AFB 1 lesions in the fetus are discussed below.

Aflatoxin induces distinctive mutational types and mutational spectra in the gpt gene of mice treated at GD14
To determine whether the types of mutations in the gpt gene were related to AFB 1 treatment and distinct from controls, gpt mutants were sequenced from both control and AFB 1treated mice (the controls had fewer mutants, as expected). The types of mutations are enumerated in Supporting Information Table 1 and included base substitutions, insertions and deletions in both AFB 1 -treated and control mice. The predominant mutations found in control liver were G:C to A:T transitions and deletions, whereas G:C to T:A transversions were most common in AFB 1 -treated animals. For example, at 10 weeks 42% of total mutants were G:C to T:A. Interestingly, while the absolute MFs of 3-week-old AFB 1treated and control mice were not statistically different (Fig.  2a), the signature G:C to T:A mutation of AFB 1 is clearly evident at 3 weeks in Supporting Information Table 1, where this mutation accounts for 49% of all mutants. Overall, at 3 and 10 weeks post birth, the B6C3F1 mice treated in utero showed a mutational pattern that is consistent with that known to be produced by the AFB 1 -N 7 -Gua and AFB 1 -FAPY adducts. 14,31 Supporting Information Figures 1 and 2, which plot the frequency distribution of types of mutations across the gpt gene, provide further granularity regarding the mutagenic properties of AFB 1 following in utero exposure. As expected, this mutational spectrum is strongly influenced by the selective conditions used to isolate 6-TG resistant mutants. Nonsense and missense mutations plus insertions and deletions producing frameshifts that inactivate gpt determine the resistant phenotype. The mutational spectrum observed for B6C3F1 mice treated in utero looks identical to that determined previously by Woo et al. 25 for animals treated at 4 days post parturition. As in the experiment done by Woo et al., we observed that the MF in liver seems to elevate by 2.5 to 3-fold between 3 and 10 weeks of age (e.g., comparison of Figs. 2a and 2b). Trying to understand the basis for this  consistently observed elevation in absolute MF over time, a program based on the Adams-Skopek algorithm for comparison of mutational spectra 32 was used to test the possibility that the distribution of mutations (and not simply the absolute MF) in the gpt gene may change between 3-and 10 weeks (Supporting Information Figs. 1 and 2). The results of the Monte Carlo based analysis indicated that the two mutational spectra were not significantly different (p 5 0.0689). Below, we speculate on other possible reasons why the MF rises between 3 and 10 weeks of age.

Developmental stage and cell proliferation in the liver
It is likely that replicating cells in the developing liver are more vulnerable than quiescent ones to genetic damage from environmental chemicals. If this sensitivity differential indeed exists, then AFB 1 -DNA adducts formed in the fetal liver would have a greater quantitative potential (on a mutation per-adduct basis) to cause mutations than AFB 1 exposure later in life. To investigate the differences in growth rate between fetal and adult livers, liver tissues were harvested and assayed for the numbers of proliferating cells by immunohistochemical staining with an antibody to Ki67. 33 Figure  4 shows that the proportion of cells expressing the Ki67 anti-gen and staining intensity varied between the developmental stages. In the fetal liver 80% of cells were Ki67 positive. In contrast, the liver specimen obtained from an adult C57BL/6J mouse had <1% of cells that stained positive for Ki67. These results provide an important proliferation metric against which to compare mutation rate and adduct density over the mouse life course. In principle, the high rate of proliferation of the fetal liver could present a substantial liability.

Discussion
Neonatal and infant mice are highly sensitive to hepatocarcinogenesis during a relatively short period following birth. 34 Intrauterine exposures to chemical carcinogens have also been demonstrated to induce liver tumors if exposures occur following organogenesis. 35 These early life periods of sensitivity are dependent upon both the metabolic capability of the liver to activate carcinogens and the rapid cell division required for growth and development. 17 In this study a quantitative association is made between AFB 1 exposures, measured as covalent DNA adducts, in adult and embryonic mice and increased mutations in offspring. The presence of AFB 1 -DNA adducts in fetal liver following administration of AFB 1 by either i.p. injection or gavage indicated that the fetal liver is able to produce the AFB 1 -exo-8,9-epoxide, the aflatoxin metabolite that best correlates with carcinogenic, toxic and mutagenic outcomes.
Evidence is provided that AFB 1 -DNA adducts formed in fetal liver DNA are much more likely to produce mutations than the same amount of DNA damage would if it occurred in the liver of an adult animal. For example, administration of AFB 1 to pregnant mice results in only about 1% the level of adducts in the fetus at 6 hr, as compared to the level in the mother. All things being equal, one would expect a mutation rate in the fetus that was similarly 1% that of the mother. The mutation rate, by contrast, was much higher than expected (Table 2), with the fetus and mother showing only a 4.6 fold difference in MFs, rather than the anticipated 100-fold difference. Taken together, our observations on longitudinal mutagenic sensitivity of mice, normalized to the level of DNA damage, indicate that AFB 1 -DNA adducts have a greater likelihood of causing mutations in the developing fetus than they do in adults.
It has long been assumed that the frequency of DNA replication correlates with the frequency of both spontaneous and DNA damage-induced mutations. We calculated an estimate of the ability of an AFB 1 adduct to produce a mutation in the gpt transgene by dividing the gpt MF by the number of adducts per megabase in liver DNA ( Table 2). The relative risk of a gpt mutation per AFB 1 adduct is 20 fold greater in the fetal liver as compared to the adult. The mitotic index of the fetal liver was at least 80 fold higher than the adult liver, which leads us to believe that the high mutation per adduct result observed is owed to the high rate of fetal liver cell division. The present data give direct experimental support to the longstanding view that the high rate of cell division of young animals helps explain their enhanced risk to DNA damage mediated disease.
In addition to a higher than expected mutagenic potency of DNA adducts in the prenatal liver, we observed that the MF of AFB 1 appears to increase, albeit slightly, as the animal matures. This observation extends a conclusion made by Woo et al. 25 where we saw that animals dosed at 4 days of age showed a slightly lower mutation frequency at 3 weeks of life compared to that at 10 weeks of life. In the present experiment, the MF during this 7-week interval (between 3 and 10 weeks of life) for mice treated in utero increased by 1.8-fold in males and 1.3fold in females (after subtraction of the MF contribution from the DMSO controls). The increased MFs were correlated with increases in liver weight from 0.8 6 0.07 to 1.4 6 0.37 g in males (1.75-fold) and from 0.72 6 0.05 to 0.97 6 0.34 g in females (1.3-fold). This correlation is consistent with the rise in mutations being a function of increased cell number occurring as the liver gains weight during this growth period. It is not immediately obvious why an increase in cell number would cause a selective increase in MF at time points so distant from the time at which the toxin was administered. One explanation for the increase in MF is the selective expansion of mutant cells. The gpt transgene present in the recoverable phage lambda vector is not expressed in the mouse and is assumed to be genetically neutral, providing no advantage or disadvantage to cell growth. 36 Consequently, it would be expected that once mutations are fixed by DNA replication, the average MF in the liver would remain constant. However, liver cell organization and function are continuously being remodeled and reprogrammed, respectively, during the late fetal and early neonatal periods resulting in unequal growth of liver cell types that could increase or decrease the observed MF. Zonal differences in cell proliferation in the liver occur during fetal development, postnatal growth and homeostasis in the adult organ. 37,38 If tissue progenitor cells are at greater risk of mutation during fetal development, mutant cells emerging from these populations would be expected to increase the MF in the tissue, which could account for the observed increase in MF at the 10-week time point.
A second factor that could contribute to the enhanced MF between 3 and 10 weeks is the possibility that "jackpot" mutations could have occurred early after dosing with AFB 1 . Statistically, however, it is unlikely that we would have observed jackpots in our analyses since the samples for mutational analysis were prepared from the entire liver. Consequently, only a small fraction of cells from a clone would be expected to be included in our analysis. Consistent with these expectations, we only rarely observed an identical gpt mutation in the same animal.
A third explanation for the increased MF during the 7-week period is the generation of new mutations arising from continued replication past AFB 1 -damage. AFB 1 -FAPY adducts, which represent 50% of AFB 1 -DNA adducts found 6 hr after dosing are very persistent. In a previous study in neonatal B6C3F1 mice the highly mutagenic AFB 1 -FAPY adduct was readily detectable in liver DNA 3 weeks after animals were dosed at 4 days of age. 31 The continued predominance of G:C to T:A transversions in the mutational spectrum at 10 weeks of age is consistent with the persistence of AFB 1 -FAPY in liver DNA after birth following transplacental exposures to AFB 1 .
One interesting observation in the present study was a higher than expected level of G:C to A:T mutations in the AFB 1 -dosed fetal liver than we observed in earlier work in an experiment in which animals of the same genotype were treated with the toxin shortly after birth. The fetal liver contains a different array of cell populations than is present in the mature liver; it is colonized by hematopoietic stem cells (HSCs) along with fetal macrophages until the end of prenatal life when HSCs migrate to the spleen and bone marrow. [39][40][41] This developmental juxtaposition places proliferating hepatoblasts in the vicinity of immune system cells capable of producing reactive oxygen species. The latter may contribute to DNA damage that is reflected in the gpt mutational spectra observed following in utero AFB 1 exposure. The majority of base substitution mutations from in utero AFB 1 exposure were G:C to T:A transversions (40-50% of gpt mutants). The next most prevalent mutation type was the G:C to A:T transition (27% of gpt mutants), which is considered an indication of base damage by oxygen radicals. 42,43 In our previous study of neonatal mice treated with AFB 1 on day 4, G:C to A:T transitions represented only 16% of total mutants. 25 This higher relative amount of G:C to A:T in the present study (27% vs. 16%) suggests the possibility that the fetus is more prone to oxidative stress.
The proximity of HSCs to hepatoblasts and other cells in the fetal liver that are capable of metabolic activation of AFB 1 could also place those cells at risk for mutation, increasing the chance of developing blood malignancies and myeloid diseases. This hypothesis is supported by the observed increase in leukemias among the F1 generation of rats after transplacental exposures to AFB 1 . 21

Implications for human disease
Our data suggest that transplacental exposures to AFB 1 , on an adduct-normalized basis, present a much greater risk of genetic change than exposures that take place later in life. One conclusion is that estimates of risk of HCC in the general population may vastly underestimate the risk of adverse effects following prenatal exposures. Risk estimates typically are based on albumin-or urinary DNA-adduct levels, which correlate with genomic damage in people. From our study, we suggest that an additional factor that takes into account the rapid cell division associated with early life is an important amplifier of risk. Because of the elevated risk of adverse health effects to offspring it is especially important to prevent maternal exposures to aflatoxins. In addition, to mitigate the potential of adverse effects due to unavoidable exposures to aflatoxins in contaminated foods, it would be useful to explore preventive measures that enhance maternal pathways of metabolism and detoxification.