Aflatoxin B1-Associated DNA Adducts Stall S Phase and Stimulate Rad51 foci in Saccharomyces cerevisiae

AFB1 is a potent recombinagen in budding yeast. AFB1 exposure induces RAD51 expression and triggers Rad53 activation in yeast cells that express human CYP1A2. It was unknown, however, when and if Rad51 foci appear. Herein, we show that Rad53 activation correlates with cell-cycle delay in yeast and the subsequent formation of Rad51 foci. In contrast to cells exposed to X-rays, in which Rad51 foci appear exclusively in G2 cells, Rad51 foci in AFB1-exposed cells can appear as soon as cells enter S phase. Although rad51 and rad4 mutants are mildly sensitive to AFB1, chronic exposure of the NER deficient rad4 cells to AFB1 leads to increased lag times, while rad4 rad51 double mutants exhibit synergistic sensitivity and do not grow when exposed to 50 μM AFB1. We suggest RAD51 functions to facilitate DNA replication after replication fork stalling or collapse in AFB1-exposed cells.


Introduction
Hepatocellular carcinoma (HCC) ranks fifth in worldwide cancer mortality (for review, see [1]) and sixth in the United States [2]. High-risk factors for HCC include exposure to genotoxins, such as the mycotoxin aflatoxin B 1 (AFB 1 ), and infection with hepatitis B and C viruses [3]. Exposure to AFB 1 is endemic in particular areas of China and sub-Saharan Africa due to Aspergillus flavus (mold) contamination of food and water [3]. A current hypothesis is that regeneration of liver cells following chronic liver injury renders liver cells susceptible to AFB 1 -associated carcinogenesis [4].
HCC pathogenesis is correlated with the accumulation of mutations and chromosomal rearrangements leading to either an inactivation of tumor suppressor genes or activation of oncogenes (for review, see [5]). MicroRNA-221 (MiR-221) overexpression contributes to liver tumorigenesis [6] and correlates with downregulation of cyclin dependent kinase inhibitors p21 and p57 [7]; however, there is no known correlation with AFB 1 exposure. The p53(Ser)249 substitution mutation frequently occurs in liver cancer, where AFB 1 exposure is highest [8][9][10]; however, there are conflicting reports whether the p53 249 codon is a direct hot spot for AFB 1 -associated mutagenesis [11]. Gross chromosomal translocations and gene amplifications have also been observed [12], and 10%-20% of HCCs contain cyclin D amplifications [13]. Although HCC associated with AFB 1 exposure exhibits more genetic instability compared to HCC in nonendemic regions [14], it is unclear which types of genetic instability are directly caused by AFB 1 -associated DNA damage. AFB 1 is not genotoxic per se but requires metabolic activation. In humans, AFB 1 metabolic activation in the liver is catalyzed by CYP1A2 and CYP3A4 [15] to form the highly unstable AFB 1 -8,9-exo-epoxide, which reacts primarily with the N 7 position of guanine, present in the major groove of DNA [16]. The resulting adduct, 8,9-dihydro-8-(N 7 -guanyl)-9-hydroxyaflatoxin B1 (AFB 1 -N 7 -Guanine) is unstable and converts to either formamidopyrimidine 2 Journal of Nucleic Acids (FAPY) derivatives or an apurinic site [16], both potentially mutagenic [17]. Both the FAPY derivatives and AFB 1 -N 7 -Guanine adducts are repaired by the nucleotide excision repair (NER) genes [18,19]. The FAPY adducts also hinder DNA replication [20], which could lead to chromosomal breaks and require DNA repair genes that function in double-strand break and X-ray repair (XRCC). Thus, repair of AFB 1 -associated DNA damage may require both NER and XRCC genes.
Interestingly, a subset of known polymorphisms [21] in both NER gene XPD and the X-ray repair gene XRCC3 correlate with higher incidence in liver cancer in endemic areas of AFB 1 exposure [22,23]. Defective NER could lead to an increase in DNA adducts, while XRCC3 polymorphisms could confer defective repair of double-strand breaks (for review, see [24]). However, the polymorphism in XRCC3, Thr241Met, which is correlated with higher levels of AFB 1 -associated HCC [23], has not been correlated with a defect in double-strand break repair [25], suggesting that other functions in DNA damage or repair may be defective in cells containing this allele. Considering that recombinational repair may also participate in DNA damagetolerance pathways, it is important to elucidate whether there are competing DNA repair pathways for AFB 1 -associated adducts.
The genotoxicity of AFB 1 in yeast [26][27][28][29] correlates with the transcription of DNA repair genes involved in recombination, including RAD51 [27,30] and RAD54 [30]. RAD51 induction has been observed in log phase cells exposed to AFB 1 [30], and RAD51 overexpression partially suppresses recombination defects in the mec1 null checkpoint mutant [27]. RAD51 is also required for AFB 1associated sister chromatid exchange (SCE) [29]. These results indicate that increased expression of RAD51 likely stimulates recombination when cells are exposed to AFB 1associated DNA adducts.
In log-phase yeast cultures, AFB 1 is a mutagen [28]. Microarray analysis reveals not only a strong induction of RAD51 and RAD54 but also a downregulation of gene transcripts encoding histones [30]. rad51 mutants exhibit an increase in AFB 1 -associated mutations [28]. Both AFB 1associated mutations and recombination events require checkpoint genes [28,29]. Considering that AFB 1 exposure triggers both checkpoint activation and S phase delay, it seems possible that the genotoxic responses to AFB 1 result from stalled replication forks.
In this paper, we show that Rad53 activation correlates with S phase delay and subsequent Rad51 foci formation in yeast. A RAD51 deletion in NER defective strains results in high levels of lethality. We suggest that Rad51 foci formation may occur in S phase and be associated with error-free bypass of DNA lesions.

Strains and Media.
Standard media were used for the culture of yeast cells. YPD (yeast extract, peptone, and dextrose), SC-TRP (synthetic complete lacking tryptophan), and SC-URA (synthetic complete lacking uracil) and FOA (5-fluro-orotic acid) were described in Burke et al. [31]. The genotypes of yeast strains used in this study are listed in Table 1. rad4, rad51 and rad4 rad51 strains for measuring SCE and AFB 1 sensitivity are derived from YB163, which contains his3 recombination substrates in tandem at TRP1 [32]. Ura − derivatives of rad4, and rad4 rad51 strains were selected on FOA medium. LSY1957 was a gift of Fung et al. [33]. pRS424-CYP1A2 was constructed by inserting the SacI CYP1A2 fragment from pCS316 into pRS424 and introduced into yeast strains by selecting for Trp + transformants.

Measuring DNA AFB 1 -Associated Recombination and
Rad51 foci. To measure AFB 1 -associated genotoxic events, log-phase yeast cells (A 600 = 0.5-1) were centrifuged and concentrated five-fold in selective media (SC-URA or SC-TRP). Cells were exposed to 50 μM AFB 1 , previously dissolved in DMSO. To synchronize cells in G1, log-phase cells were exposed to 10 −4 M alpha factor (Sigma Co.) for two hours, and G1 cells were visualized in the light microscope. Cells were maintained in selective media (SC-URA or SC-TRP) during the carcinogen exposure and then washed twice in H 2 O. To measure SCE frequencies, recombinants were selected on SC-HIS, and viability was determined by plating an appropriate dilution on YPD. To visualize Rad51 foci formation, cells were resuspended in selective media (SC-TRP or SC-URA) and immobilized on glass slides.
To determine whether ionizing radiation stimulates the formation of Rad51 foci, cells were washed once in H 2 O, resuspended in 10 ml of H 2 O, and placed in a 81 mm diameter Petri dish. Cells were irradiated at 6 krad using a Nordion 1.8 kCi 137 Cs irradiator (6 krad/hr). After irradiation cells were concentrated in YPD medium and immobilized on glass slides.

Live Cell Epifluorecence and Microscopy Analysis.
Cells for microscopic analysis were grown to early-mid-log phase overnight in synthetic medium. After exposure to the genotoxin, cells were harvested by centrifugation, washed twice, and resuspended in YPD. Immobilization of cells was performed by mixing equal volumes of cell suspension and 1.4% low-melt agarose. Cover slips were sealed with a wax mixture as described by Lisby et al. [35]. Slides were visualized using a Zeiss LSM 510 META confocal microscope.

FACS Analyses.
Cells were visualized by the fluorescent activated cell sorter (FACS) to directly correlate their DNA content with their cell-cycle phases. After AFB 1 exposure, cells were washed, resuspended in 0.1 M sodium citrate, and treated with 1 mg/ml RNase A at 50 • C for 1 hr. The cells were incubated at 50 • C for 5 hr in 8 μg/ml proteinase K. An equal volume of 25 μg/ml propidium iodide (PI) diluted in 0.1 M sodium citrate was then added to the cells prior to the FACS analysis, to a final concentration of 12.5 μg/ml PI, and analyzed for fluorescence content with the use of a BD LSR II Flow cytometer.

Detection and Quantification of DNA Adducts.
To measure the AFB 1 -associated DNA adducts in yeast, we used liquid chromatography-electrospray ionization tandem mass spectrometry (LC-ESI/MS/MS) [36]. Log-phase cultures of yeast-expressing human CYP1A2 (pCS316) were exposed to 50 μM AFB 1 for 4 h. Because standard protocols for isolating yeast DNA involve alkaline buffers, rendering the highly unstable AFB 1 -N 7 -Guanine DNA adducts labile, we have modified the "smash-and-grab" protocol [37] so that we are using a neutral buffer containing 10 mM Tris-HCl, 1 mM EDTA, 100 mM NaCl, 2% Triton X-100, and 1% SDS, pH 7. DNA was isolated from two independent samples of yeast cells. The DNA adducts were identified and measured by high-performance liquid chromatography and tandem mass spectroscopy (LC-ESI/MS/MS) after acid hydrolysis [36].
2.6. Determining Rad53 Activation. Activation of Rad53 was determined by Western blots. Cells were inoculated in SC-URA medium. Log-phase cells (A 600 = 0.5-1) were concentrated three fold in SC-URA and exposed to 50 μM AFB 1 for 4 h. After washing cells twice in H 2 O, aliquots were plated directly on SC-HIS to measure recombination, appropriately diluted and plated on YPD to measure viability. Protein extracts were prepared as previously described by Foiani et al. [38], separated on 10% acrylamide/0.266% bis-acrylamide gels for Rad53 detection and transferred to nitrocellulose membranes. Rad53 was detected by Western blotting using goat anti-Rad53 (yC-19, Santa Cruz, Biotechnology, Santa Cruz, CA). The secondary antibody was antigoat IgG-HRP (Santa Cruz).

Delay in Cell-Cycle Progression Correlated with Rad53
Activation. We previously observed that log-phase cells exposed to AFB 1 exhibit Rad53 activation [29], which can result from replication blockage or delay. We measured Rad53 activation and cell-cycle delay in log-phase cells after continuous exposure to 50 μM AFB 1 . The data show that there is a peak activation of Rad53 after 3 hrs exposure to AFB 1 . Three hrs of exposure was also sufficient time to observe SCE recombinants (Figure 1). After 4 hrs of AFB 1 exposure, there was less Rad53 activation and cell-cycle progression continued. The transient delay in the cell cycle is consistent with a previous study that indicated that AFB 1exposed yeast exhibit a transient S phase delay [30]. The data suggest that there is a correlation between AFB 1 -associated Rad53 activation and S phase delay. Because AFB 1 adducts are detected in cycling cells after the S phase delay [29], we speculate that cells actively tolerate AFB 1 -associated DNA lesions during DNA replication.
Rad53 not only delays cell-cycle progression but is also required for the phosphorylation of the Rad51 paralogs, Rad55, and Rad57, which may facilitate replication restart at stalled replication forks [39]. Rad55 and Rad57 facilitate the formation of DNA repair foci at double-strand breaks [40]. Previous data indicate that RAD53 is required for DNA damage associated SCE [41], including AFB 1 -associated SCE [29]. We, therefore, determined whether AFB 1 exposure also stimulates Rad51 foci formation in yeast. 1 Results in Rad51 foci Formation in Cells that Are Entering S Phase. To visualize Rad51 foci that result from AFB 1 -associated DNA damage, we introduced pRS424-CYP1A2 into strain LSY1957 [33] to metabolically activate AFB 1 . This strain (YB405) contains rad51-I345T, an allele of RAD51, which when tagged with YFP, is still capable of conferring radiation resistance [33]. As a positive control, cells were exposed to either X rays (2 krad) or gamma rays (6 krads). After exposure, cells were returned to growth medium (YPD), and live cells were imaged for Rad51 foci. After 2 hrs of growth in YPD, most irradiated cells arrested in G2 and exhibited the dumb-bell shape (Figure 2). Cells were then visualized with the confocal microscope. Nearly 90% of    Figure 1: Rad53 checkpoint activation, cell cycle progression, and recombination after log-phase cells were exposed to 50 μM AFB 1 . At indicated times, cells were collected for FACS analysis to measure frequencies of SCE and to make extracts to measure Rad53 activations. (a) Rad53 checkpoint activation was monitored after AFB 1 exposure at the indicated times. Rad53 and the activated checkpoint protein, Rad53p, are indicated by arrows. (b) FACS analysis was performed at indicated time periods after exposure. The G1 and G2 cells are indicated by P4 and P5. The peak to the right of the G2 peak indicates bloated cells due to enlarged vacuoles. (c) AFB 1 -associated SCE were measured by selecting for His + recombinants that result from unequal recombination between two truncated his3 fragments. Net recombination frequencies = recombination frequency after AFB 1 exposure-spontaneous recombination frequency. the G2-arrested cells contain Rad51 foci, in agreement with Lisby et al. [40].

Exposure to AFB
Similarly, we determined whether Rad51 foci appear in cells after exposure to 50 μM AFB 1 for 3 hrs. To first confirm the presence of AFB 1 adducts, we extracted DNA from LSY197 cells after AFB 1 exposure and observed approx-imately the same number of DNA adducts (Table 2) as previously observed in strains used to measure recombination [29]. After 3 hr of AFB 1 exposure, we also observed that nearly 90% of the cells exhibited Rad51 foci. However, the difference with irradiated cells was that many AFB 1 -exposed cells that exhibited Rad51 foci were not G2 arrested. In Journal of Nucleic Acids Figure 2: Confocal imaging of Rad51-YFP foci in live AFB 1 -exposed yeast cells. (a) Control, no AFB 1 exposure, (b) Rad51-foci after exposure to X-rays, (c) Rad51 foci in logarithmically growing cells exposed to 50 μM AFB 1 , wide view, and (d) Rad51 focus in a synchronized cell after alpha factor arrest and exposure to AFB 1 .
addition, many cells exhibited Rad51 foci in both mother and daughter bud ( Figure 2). These data indicate that AFB 1associated Rad51 foci are not restricted to the G2 phase of the cell cycle. Because daughter buds are not always visible in the confocal microscope, we decided to synchronize cells in G1, expose the cells to AFB 1 and then determine when Rad51 foci could be detected.
To determine whether cells can express Rad51 foci in S phase, cells were first arrested in G1 with alpha factor, and then exposed to AFB 1 for three hours. Cells were then washed and returned to growth medium without AFB 1 . We observed that newly budded cells (90%) contain Rad51 foci. Rad51 foci were not evident after 30 minutes or 1 hr incubation time, but were evident after 1.5 hrs of incubation; after three hours of incubation, there were few Rad51 foci. These data indicate that Rad51 foci can be observed in cycling cells that are entering S phase.  Table 1 for complete genotype, b n = 2, c Ratio: AFB 1 adduct in strain or rad mutant/AFB 1 adduct in wild type (YB163).

rad4 Cells
Are Defective in the Excision of AFB 1 DNA Adducts. NER and recombinational repair mutants are modestly sensitive to AFB 1 [27,28]. Both rad4 and rad51 mutants exhibit higher rates of AFB 1 -associated mutagenesis [29,30]. Measurements of DNA adducts indicated there are about three-fold higher levels of AFB 1 -N 7 -Guanine adducts in rad4, compared to wild type (Table 2). Consistent with the notion that AFB 1 adducts persist longer in rad4 mutants, we observed by FACS analysis a delayed S phase after exposure to 10 μM AFB 1 . We asked whether wild type, rad4, and rad51 mutants could tolerate DNA adducts. We used a growth assay in microtiter dishes [42] to determine differences in growth curves of wild type (YB401), rad4 (YB403), rad51(YB402), and rad4 rad51 (YB404) strains during chronic exposure to AFB 1 . Approximately 10 5 cells were inoculated in 96 well plates and exposed to 0 μM, 25 μM and 50 μM of AFB 1 ; the experiments were done in triplicate. Growth was measured by A 600 (Figure 3). The lag time [42] for wild type was ∼3 hrs and similar to rad51. Whereas the rad4 mutants exhibited a longer lag period, ∼13 hrs, the rad4 rad51 mutant exhibited little, if any, growth. The data suggest that RAD51 function is critical in conferring AFB 1 resistance in the rad4 mutant. This result is consistent with previous results that rad14 rad51 cells are also synergistically more sensitive to AFB 1 [30].
To further investigate whether rad4 cells can progress through S phase in the presence of DNA adducts, we arrested rad4 cells in G1 with alpha factor and exposed cells to AFB 1 for 3 hrs. Cells were then washed, diluted, and inoculated on YPD plates to visualize the growth of single colonies every three hours under the microscope. After 12 hrs, ∼70% (46/67) of cells that were not exposed to AFB 1 formed colonies. However, only ∼16% (46/268) of cells exposed to AFB 1 (10 μor 50 μ) formed colonies. 60% (n = 2) of either wild type or rad4 cells that do form colonies retain the URA3-containing plasmid expressing CYP1A2, indicating that colony formation occurred in cells that could still metabolically activate AFB 1 . Many of the rad4 cells that do not form colonies were arrested in early S phase, as evident by the presence of small daughter buds (Figure 3(f)). These data indicate that a few NER-deficient cells can progress through the cell cycle in the presence of AFB 1 -associated DNA adducts.

Discussion
AFB 1 is a very potent liver carcinogen. The metabolic activation of AFB 1 generates AFB 1 -associated DNA adducts which both stimulate mutagenesis and recombination in a variety of organisms. Polymorphisms in both XPD and XRCC3 are correlated with greater HCC risk [22,23], thus underscoring the need to elucidate the role of recombinational repair in AFB 1 metabolism. In budding yeast, well-conserved RAD and checkpoint genes are required for AFB 1 -associated mutation and recombination [28,29]. Higher frequencies of AFB 1 -associated mutations in rad51 mutants [28,29] and synergistic increase in the AFB 1 sensitivity of double mutants defective in both recombinational and NER suggest that there are redundant pathways in conferring resistance to AFB 1 . AFB 1 is highly recombinogenic and induces RAD51 in yeast [27]; however, it was unclear when and if Rad51 foci form. The salient conclusions from this study are (1) cells are delayed in S phase after AFB 1 exposure, and there is a correlation with Rad53 checkpoint activation and cell cycle delay; (2) Rad51 foci form after AFB 1 -exposed cells enter S phase; (3) rad4 mutants accumulate AFB 1 -N 7 -Guanine adducts, and many exposed cells arrest or are delayed in S phase. This is the first study to show that AFB 1 exposure leads to Rad51 foci formation.
Previous studies indicated that RAD53 is required for both AFB 1 -associated recombination and mutagenesis in yeast [29]. The coincident timing of recombination, Rad53 checkpoint activation, and S phase delay suggest that checkpoint activation is required to trigger AFB 1 -associated recombination in yeast. A possible link is that Rad53 is required for the DNA damage-associated phosphorylation of the Rad51 paralogs Rad55 and Rad57 [39], which participate in recombinational repair between sister chromatids [34]. We do not yet know how the checkpoint activation would trigger mutagenesis, but several studies suggest a role for checkpoint genes in DNA damage tolerance and translesion synthesis [43][44][45][46].
The identification of AFB 1 -associated Rad51 foci was performed by detecting YFP fluorescence in the confocal microscope. Rad51 is not known to bind to specific DNA adducts, so we presume that the AFB 1 N 7 -Guanine adducts and resulting FAPY and apurinic sites are further converted into recombinogenic lesions, including double-strand breaks or single-strand gaps. It is unlikely that all the lesions that initiate AFB 1 -associated Rad51 foci are the same as for Xray-associated Rad51 foci, since we observed Rad51 foci in newly cycling cells entering S phase whereas we only observed X-ray-initiated Rad51-foci in G2-arrested cells. However, double-strand breaks or single-strand gaps could also result after replication forks stall or collapse in S phase, and AFB 1 lesions have been reported to stall or hinder DNA replication in Escherichia coli [17]. Thus, an attractive model is that Rad51 foci form as AFB 1 -exposed cells transition through S phase.
We suggest that RAD51 confers AFB 1 resistance in NER defective mutants by two possible functions. First, RAD51 would function in repairing double-strand breaks that indirectly results from AFB 1 -associated DNA damage. Considering that one double-strand break could confer lethality [47], one would estimate that at least one break occurs in every rad4 rad51 double mutant cell during chronic exposure to 50 μM AFB 1 . Second, RAD51 could actively participate in tolerating AFB 1 -associated DNA lesions; previous studies have indicated that the RAD52 pathway is involved in tolerating UV-associated damage [48]. Growth curves of wildtype cells indicate that some AFB 1 -associated DNA adducts can be tolerated without affecting doubling time. This second possibility is supported by observations that rad14 rad51 mutants exhibit extremely high frequencies of AFB 1associated mutagenesis [28]. Further speculation would suggest that the Rad51-paralog, XRCC3, has a similar function.  Table 1, for complete genotype). 10 5 log-phase cells were inoculated in each well, n = 3. A 600 is plotted against time (h).

Conclusions
(Right) rad4 cells synchronized in G1 were exposed to 50 μM AFB 1 for 3 hrs, washed, and inoculated on YPD. Cell growth was monitored from 0-18 hrs. Bars indicate the actual range of measurements.
Rad51 foci during the cell cycle in early S phase suggests a different mechanism of foci formation, compared to that of ionizing radiation. Understanding the function of these Rad51 foci will elucidate how polymorphisms in XRCC3 correlate with higher rates of liver cancer in endemic areas exposed to AFB 1 . It will thus be interesting if Rad51 foci also occur in mammalian cells after AFB 1 exposure.