DNA fragile site breakage as a measure of chemical exposure and predictor of individual susceptibility to form oncogenic rearrangements

Summary

Non-cytotoxic levels of environmental/chemotherapeutic agents preferentially induce breakage in fragile site genes including RET. Further, blunt-ended DNA breaks at RET are increased in normal cells of patients with RET/PTC-driven tumors, and this suggests susceptibility of rearrangement formation and implicates NHEJ.

Abstract

Chromosomal rearrangements induced by non-radiation causes contribution to the majority of oncogenic fusions found in cancer. Treatment of human thyroid cells with fragile site-inducing laboratory chemicals can cause preferential DNA breakage at the RET gene and generate the RET/PTC1 rearrangement, a common driver mutation in papillary thyroid carcinomas (PTC). Here, we demonstrate that treatment with non-cytotoxic levels of environmental chemicals (benzene and diethylnitrosamine) or chemotherapeutic agents (etoposide and doxorubicin) generates significant DNA breakage within RET at levels similar to those generated by fragile site-inducing laboratory chemicals. This suggests that chronic exposure to these chemicals plays a role in the formation of non-radiation associated RET/PTC rearrangements. We also investigated whether the sensitivity of the fragile RET region could predict the likelihood of rearrangement formation using normal thyroid tissues from patients with and without RET/PTC rearrangements. We found that normal cells of patients with thyroid cancer driven by RET/PTC rearrangements have significantly higher blunt-ended, double-stranded DNA breaks at RET than those of patients without RET/PTC rearrangements. This sensitivity of a cancer driver gene suggests for the first time that a DNA breakage test at the RET region could be utilized to evaluate susceptibility to RET/PTC formation. Further, the significant increase of blunt-ended, double-stranded DNA breaks, but not other types of DNA breaks, in normal cells from patients with RET/PTC-driven tumors suggests that blunt-ended double-stranded DNA breaks are a preferred substrate for rearrangement formation, and implicate involvement of the non-homologous end joining pathway in the formation of RET/PTC rearrangements.

Introduction

Chromosomal rearrangements are a common genetic abnormality involved in the initiation of cancer development. These rearrangements result in the disruption of genetic material, which can lead to the expression of oncogenic fusion proteins or the disruption of processes involved in tumor suppression (1). DNA strand breaks must occur in participating gene regions to initiate all chromosomal rearrangements. We have shown that fragile site-inducing conditions can create DNA breaks within RET/PTC participating genes and ultimately lead to the formation of RET/PTC rearrangements (2). RET/PTC rearrangements are a common mutation observed in papillary thyroid carcinoma (PTC) (3) accounting for approximately 20% of adult (3) and approximately 45% of childhood sporadic PTC cases (4). PTC is responsible for the rising incidence of thyroid cancer in the USA (5), where thyroid cancer is now the most rapidly increasing cancer type (5,6). Although increases in thyroid cancer have been well documented following exposure to high doses of radiation (7,8), many tumors are sporadic in nature (4,9), suggesting other factors are involved in initiating DNA breaks in RET/PTC genes. RET, an oncogene involved in recurrent chromosomal rearrangements found in thyroid (3,4,9,10) and lungs (11,12) is located within fragile site FRA10G. Our previous study offers direct evidence for the role of fragile sites in cancer-specific rearrangements (2).

DNA fragile sites are sensitive to a variety of chemicals and have been identified in regions with deletions and chromosomal rearrangements (13,14). Environmental agents such as benzene (in cigarette smoke and automobile exhaust) and diethylnitrosamine (DEN) (in cigarette smoke, pesticides and cured meat), and chemotherapeutic drugs, can significantly increase fragile site breakage (13,14) and are positively associated with the risk of thyroid cancer (15–26). Benzene has been associated with an increased risk of thyroid cancer after exposure to volcanic activity at Mount Etna (19), and an increased risk of thyroid cancer was observed in female textile workers with 10 or more years of benzene exposure (25). Specific nitrosamines, including DEN, have been shown to cause thyroid tumors in animal studies (27). Increasing dietary nitrate intake (which generates nitrosamine by interacting with amines in vivo, or by cooking) is positively associated with thyroid cancer risk in older women (24) and men (17). Chemotherapy regimens which contain the topoisomerase II inhibitors etoposide or doxorubicin have been linked to secondary PTC following treatment for primary osteosarcoma (18), rhabdomyosarcoma (22), Ewing’s sarcoma and Wilms tumor (15). We have found that DNA topoisomerases I and II participate in initiating fragile site breakage at the RET oncogene (28). Therefore, investigating the sensitivity of the RET region to these chemicals is critical for understanding whether these agents are likely to contribute to the formation of non-radiation induced RET/PTC rearrangements, which are more common than radiation-induced RET/PTC in thyroid cancer.

While common fragile sites are found in all individuals, the extent of DNA breaks at particular fragile sites varies among individuals (29). Double-stranded DNA breaks are commonly repaired through homologous recombination or non-homologous end joining (NHEJ) (30), but dysfunction in these pathways can contribute to increased DNA breakage. Alternatively, an accumulation of chemical exposures could increase the number of DNA breaks and overwhelm cellular DNA repair pathways, resulting in the formation of chromosomal rearrangements (1). Therefore, the DNA fragility at rearrangement-participating and breakage-sensitive gene regions may be indicative of an individual’s exposure to environmental factors and/or an unfavorable genotype for DNA stability, and thus predict a potential for the formation of cancer-causing chromosomal rearrangements. Both canonical NHEJ and microhomology-mediated end-joining (MMEJ or alt-NHEJ) are implicated in the formation of inter- and intra-chromosomal rearrangements (31–35). Studies by Ghezraoui et al. (36) demonstrated that chromosomal translocations depend on canonical NHEJ in human cells, and the ends of DNA breaks, such as blunt or long staggered, can influence the characteristics of rearrangement fusion points.

In this study, we first demonstrated that RET intron 11, the major patient break point region found in PTC patients (37–40) is sensitive to DNA breakage when exposed to non-cytotoxic, low doses of environmental chemicals (benzene or DEN) or chemotherapeutic agents (etoposide or doxorubicin). This sensitivity was also detected in the form of blunt-ended DNA double-stranded breaks generated by these exposures. Using normal human thyroid tissue samples from patients with or without RET/PTC rearrangements, we found that patients with thyroid cancer driven by RET/PTC rearrangements have significantly more blunt-ended double-stranded breaks at RET intron 11 compared to patients without the rearrangements, suggesting that this increased breakage leads to the formation of RET/PTC rearrangements. The results support the hypothesis that chronic low-dose exposure to environmental fragile site-inducing chemicals or residual chemotherapy agents contributes to the formation of non-radiation induced RET/PTC rearrangements, which are more common than radiation-induced RET/PTC in thyroid cancer. Also, the significant increase in blunt-ended DNA breaks within RET in normal cells of RET/PTC patients implicate blunt-ended DNA breaks in the formation of RET/PTC rearrangements. To our knowledge, this is the first study to demonstrate the sensitivity of a cancer driver gene in normal cells, suggesting that a test to detect DNA breakage at RET could provide an early indication of susceptibility to PTC.

Materials and methods

Cell line and culture conditions

Human thyroid epithelial cells immortalized by SV-40 (HTori-3) were purchased from the European Collection of Authenticated Cell Culture and were grown in RPMI1640 medium (Gibco) supplemented with 10% fetal bovine serum and penicillin/streptomycin.

Tissue sample procurement

Tissue samples were obtained from the University of Pittsburgh Health Sciences Tissue Bank using a protocol approved by the University of Pittsburgh Institutional Review Board. Two paired samples (tumor and normal surrounding tissue) were obtained from each of six patients with an RET/PTC+ tumor. Seven samples of normal thyroid tissue were also obtained from patients with benign nodules but without PTC (PTC−). Prior to procurement for this study, tissue samples were histologically analyzed to determine tumor tissue and surrounding normal margins. Further, reverse transcriptase-polymerase chain reaction (RT-PCR) as described previously (2), was used to verify the presence of RET/PTC rearrangements in each tumor sample and to confirm lack of RET/PTC rearrangements in the surrounding normal tissue or in the normal tissue provided by PTC− patients.

Cell treatments and analyses

Cell viability was assessed as previously described (28) using propidium iodide staining and quantification by flow cytometry. Based on known fragile site-inducing conditions (13), HTori-3 cells were treated for 24 h with 0.4 µM aphidicolin (APH) and various concentrations of benzene (0.5–1 mg/ml; Sigma-Aldrich), DEN (3.5–7 mg/ml; Sigma-Aldrich), etoposide (0.3–0.5 µM; Sigma-Aldrich) or doxorubicin (5–10 nM; Sigma-Aldrich), to determine the optimal dosage which did not alter cell viability.

To analyze active apoptosis, HTori-3 cells were treated with the chemicals described above for 24 h, harvested and resuspended in 1X Annexin V binding buffer containing Annexin V (BD Biosciences). Early apoptotic cells were then quantified using a FACSCalibur flow cytometer and FlowJo software.

To analyze cell growth, the number of viable HTori-3 cells was determined by trypan blue exclusion using a hemocytometer. Cells were plated at 1 × 10⁵ viable cells and treated with the indicated chemical concentrations. After 24 h of chemical treatment, viability was determined and the cells were washed with PBS and re-plated in chemical-free media to recover for an additional 24 h before being quantified again.

For breakpoint detection, HTori-3 cells (1 × 10⁵) were plated in six-well plates and treated approximately 18 h later with 0.4 µM APH, 0.5 mg/ml benzene, 3.5 mg/ml DEN, 0.3 µM etoposide or 5 nM doxorubicin for 24 h. Genomic DNA was then isolated from treated and untreated cells in parallel for DNA break analyses.

Detection and quantification of all DNA breaks by ligation-mediated PCR (LM-PCR)

The total number of DNA breaks including single-stranded nicks and double-stranded DNA breaks was detected and quantified as described previously (2,28) (Supplementary Figure 1A, left panel). Briefly, genomic DNA was isolated from HTori-3 cells or normal thyroid tissues of patients, followed by primer extension using a 5′-biotinylated primer matched to the regions of interest. DNA breaks were ligated to the asymmetrical duplex LL3/LP2 linker and then isolated using streptavidin beads. Amplification of the DNA breaks was achieved through nested PCR of the extension-ligation products for each region of interest. Linker and primer sequences have been described previously (2). PCR products were resolved by gel electrophoresis and sequenced to verify their identity. Each band visualized on the gel represents one break isolated within the gene region of interest. DNA breaks were then quantified as the number of DNA breaks per 100 cells.

Detection and quantification of double-stranded DNA breaks by LM-PCR

To detect only blunt-ended double-stranded DNA breaks, genomic DNA isolated from HTori-3 cells or normal thyroid tissues of patients was subjected to a protocol described previously (41) (Supplementary Figure 1A, right panel). Genomic DNA was directly ligated to the asymmetrical duplex LL3/LP2 linker, and ligation mixtures were purified through Sephadex G-100 columns prior to nested PCR. PCR products were resolved by gel electrophoresis, sequenced and quantified as described above.

To detect all double-stranded DNA breaks, genomic DNA was first incubated with E. coli DNA polymerase I large (Klenow) fragment (New England Biolabs) followed by heat inactivation at 75°C. The Klenow fragment converts double-stranded DNA breaks with either 5′- or 3′-overhangs to blunt ends through its polymerase and 3′-5′-exonuclease activities. The blunt-ended DNAs were then ligated directly to the LL3/LP2 linker. Detection and quantification of the DNA breaks was performed as described above.

To detect the combination of 5′-overhang and blunt-ended double-stranded DNA breaks, genomic DNA was incubated with the Klenow Fragment (3′–5′ exo-) lacking 3′-5′-exonuclease activity (New England Biolabs), which creates blunt-ended DNA only from 5′-overhang DNA breaks. Detection and quantification of the DNA breaks was performed as described previously.

To verify the ability of the assay to detect the specified type of DNA break, genomic DNA from HTori-3 cells was digested with various restriction enzymes (New England Biolabs) to generate either single-stranded nicks (by Nt.BstNBI digestion), 5′-overhangs (BbvI digestion), 3′-overhangs (BanII digestion) or blunt-ended (HaeIII digestion) double-stranded DNA breaks, and subjected to the four protocols listed in Supplementary Figure 1A, available at Carcinogenesis Online.

Analysis of individual types of DNA breaks

Normal thyroid tissue from patients with RET/PTC rearrangements or patients with benign nodules but without PTC, was evaluated individually for each types of breaks as follows. To determine the amount of single-stranded nicks only, the number of all double-stranded DNA breaks (from Klenow fragment-treated DNAs) was subtracted from the number of all DNA breaks. To quantify 3′-overhang DNA breaks only, the number of 5′-overhang and blunt-ended double-stranded DNA breaks (from 3′ to 5′ exo-treated DNA) was subtracted from the number of all double-stranded DNA breaks (from Klenow fragment-treated DNAs). The amount of 5′-overhang DNA breaks was quantified by subtracting the number of blunt-ended double-stranded DNA breaks with the number of 5′-overhang and blunt-ended double-stranded DNA breaks (from 3′ to 5′ exo-treated DNA).

Results

Environmental chemicals and chemotherapeutic drugs induce significant DNA breakage within fragile site genes including RET intron 11

Intron 11 of the RET gene, the major patient break point region (37–40), is sensitive to low doses of APH, a classic fragile site-inducing condition (2). DNA breakage at this region, under fragile site-inducing conditions, leads to the formation of RET/PTC rearrangements (2). Many fragile site-inducing chemicals such as benzene and DEN are encountered daily and are associated with an increased risk of thyroid cancer (17,24,25). In order to mimic daily, low level exposure to environmental chemicals, we used a cell viability assay to determine optimal, non-cytotoxic dosages of benzene and DEN to ensure that significant levels of cell death were not induced relative to the untreated control (Figure 1A). Treatment of human thyroid epithelial cells (HTori-3) with 0.5 mg/ml benzene or 3.5 mg/ml DEN for 24 h did not induce apoptosis, as measured by flow cytometry analysis of annexin V staining (Figure 1B). Cell growth, measured by trypan blue exclusion over a period of 48 h, was not perturbed following the same chemical treatment conditions (Figure 1C).

Figure 1.

Cell survival of HTori-3 following treatment with APH, benzene, DEN, etoposide or doxorubicin. (A) Cell viability following 24-h chemical treatment was determined by propidium-iodide (PI) stain and measured using flow cytometry. The percentage of live cells (PI negative) relative to untreated was determined for each chemical treatment. (B) The number of HTori-3 cells undergoing active apoptosis following 24-h chemical treatment was determined using Annexin V stain and measured by flow cytometry. The percentage of non-apoptotic cells (Annexin negative) relative to untreated is depicted. (C) The amount of cell growth following treatment was quantified using a hemocytometer. Cells were counted at the time of plating (0 h), after 24-h treatment (24 h) or allowed to recover for an additional 24 h after chemicals were removed (48 h). All data are represented as the mean ± standard deviation of at least three replicated experiments. Asterisks indicate P < 0.01 as compared to untreated following analysis by one-way ANOVA and Dunnett’s multiple comparison test.

Therefore, these conditions were used to determine whether RET intron 11 was sensitive to DNA breakage upon exposure to non-cytotoxic levels of benzene or DEN. HTori-3 cells were treated with 0.5 mg/ml benzene, 3.5 mg/ml DEN or 0.4 µM APH (as a positive control) for 24 h before being harvested for LM-PCR analysis of all types of DNA breaks, including single-stranded nicks and double-stranded DNA breaks with 5′-overhangs, 3′-overhangs and blunt ends (Supplementary Figure 1A, available at Carcinogenesis Online, left panel (2);). Both benzene and DEN induced significantly higher DNA breakage at RET compared to untreated cells (Figure 2A and Supplementary Figure 1B, available at Carcinogenesis Online). Also, the frequencies of DNA breakage induced by either of these chemicals were similar to that induced by APH, which we have shown leads to the formation of RET/PTC rearrangements (2). FHIT, encompassing the fragile site FRA3B (42), also showed a significant increase in breakage upon treatment with either chemical, while the non-common fragile site, G6PD region (2), was insensitive to fragile site induction by these environmental chemicals as well as APH (Figure 2A).

Figure 2.

Frequency of DNA breakage within RET, FHIT and G6PD regions in HTori-3 cells after 24 h treatment with APH, benzene, DEN, etoposide or doxorubicin. (A) The breakage frequency is represented as all DNA breaks per 100 cells which includes single-stranded nicks, 5′-overhang, 3′ overhang and blunt-ended double-stranded DNA breaks. (B) The frequency of only blunt-ended double-stranded DNA breaks per 100 cells after 24-h treatment. Both assays are represented by the mean ± SEM of at least three replicated experiments. Asterisks indicate P < 0.05 as compared to the respective untreated sample following analysis by one-way Anova and Dunnett’s multiple comparison test.

Next, we investigated the induction of blunt-ended double-stranded breaks within RET intron 11 by these chemicals, since this type of break could be an immediate precursor in the formation of rearrangements. Genomic DNA from treated or untreated cells was subjected to a modified LM-PCR procedure detecting only blunt-ended breaks (Supplementary Figure 1A, available at Carcinogenesis Online, right panel). Treatment with either benzene or DEN significantly increased the number of blunt-ended double-stranded DNA breaks in this region as compared to untreated cells (Figure 2B). Further, blunt-ended double-stranded DNA breaks within FHIT were also significantly increased compared to untreated cells, but with no difference in the G6PD region. This result suggests that benzene and DEN induce fragile site-specific breakage within human thyroid epithelial cells and demonstrate the sensitivity of human thyroid epithelial cells to low-dose chemical exposures. This induction of DNA breakage at RET further suggests a role for these long-term, low-dose chemical exposures in the formation of non-radiation induced RET/PTC rearrangements, which are more common than radiation-induced RET/PTC in thyroid cancer.

Treatment of primary tumors with chemotherapy regimens including topoisomerase II inhibitors such as etoposide or doxorubicin has been previously linked to secondary cancers including PTC (15,18,22). Using previous pharmacokinetic studies to mimic residual low doses following chemotherapy administration (43,44), we treated HTori-3 cells with various concentrations of each drug for 24 hours and analyzed cell viability, apoptosis and cell growth. Figure 1 shows that 0.3 µM etoposide or 5 nM doxorubicin did not induce significant cell death or apoptosis, and cell growth was not hindered.

Next, we examined the overall DNA breaks in HTori-3 cells after 24 h treatment with 0.3 µM etoposide or 5 nM doxorubicin. Both chemicals significantly increased breakage within intron 11 of RET as well as in FHIT (Figure 2A and Supplementary Figure 1B, available at Carcinogenesis Online) even at these low non-cytotoxic doses. Treatment with either drug, however, did not induce significant breakage within G6PD, again suggesting the specific sensitivity of DNA fragile sites to these drugs. Importantly, these low dose concentrations are approximately 300–1000-fold lower than the concentrations found in the plasma of patients immediately after chemotherapy treatment (43,44), demonstrating the deleterious consequences to cells that are not killed by treatment with chemotherapeutic agents.

Further, we specifically examined blunt-ended double-stranded DNA breakage induced after exposure to either chemical. The results show that these low-dose chemotherapy drugs also induce significant blunt-ended DNA breakage within intron 11 of RET and within FHIT (Figure 2B). This suggests that cells are still susceptible to DNA damage in these regions when exposed to a residual dose of chemotherapy treatment, and the significant amount of blunt-ended double-stranded breaks could serve as a direct substrate to generate cancer-specific rearrangements. Therefore, DNA breakage at these sensitive fragile regions may be a useful metric for patient prognosis and the risk of developing therapy-related secondary cancers.

Frequency of RET breakage in RET/PTC patient normal tissues is predictive of RET/PTC rearrangements

The presence of RET/PTC rearrangements in patients with RET/PTC-driven thyroid cancer suggests prior chemical and environmental agent exposures and/or unfavorable genotypes for DNA stability, which cause breakage in the intron 11 region of RET. Therefore, we hypothesize that the sensitivity of RET intron 11 signals the net balance of an individual’s exposure to environmental factors and unfavorable genotypes, and can predict the likelihood of RET/PTC rearrangements. If indeed RET breakage can indicate a potential to form the rearrangements, normal cells of RET/PTC patients should have higher DNA breakage at RET than patients without the rearrangements.

To test this, normal thyroid tissue surrounding RET/PTC+ tumors was compared with normal thyroid tissue from patients with benign nodules. First, total RNA was isolated from each tumor tissue to verify the presence of RET/PTC rearrangement in the tumor. To ensure a lack of significant tumor cell contamination within the normal tissues, total RNA was isolated from normal tissue of six patients with RET/PTC rearrangements and seven patients with benign nodules, and a lack of RET/PTC rearrangement in these thyroid tissues was verified by RT-PCR (data not shown). Genomic DNA was also isolated from each normal tissue, and DNA breakage within RET was evaluated.

Using the procedure to detect only blunt-ended, double-stranded DNA breakage, we found that normal cells surrounding RET/PTC+ tumors have a significantly increased frequency of blunt-ended DNA breaks within RET intron 11, as compared to the normal thyroid cells from patients without RET/PTC (P < 0.001) (Figure 3A and Supplementary Figure 1C, available at Carcinogenesis Online). This significant increase in DNA breakage was specific to RET intron 11, as there were no significant differences in the breakage frequency within FHIT, ALK or two non-CFS control regions, 12p12.3 or G6PD (45). ALK, located within the fragile site FRA2N, is involved in rearrangements found in up to 10% of thyroid cancer cases (46), therefore, the lack of a significant difference in the ALK region between our patient groups demonstrates the specificity of RET/PTC rearrangement formation following increased DNA breakage at RET. Interestingly, normal cells from patients with RET/PTC1 driven tumors have a significant increase in DNA breakage at the partner gene, CCDC6, compared to PTC− patient tissues (Supplementary Figure 2, available at Carcinogenesis Online). Also, no difference in DNA breaks was observed between these two sample groups at the FHIT/FRA3B region or the ALK/FRA2N, indicating that individuals susceptible to RET/PTC rearrangements might not have globally susceptible fragile sites, but have fragile site breaks specific to the individual’s translocation. These results suggest that DNA breakage at specific fragile sites could be a valuable indicator for the potential formation of chromosomal rearrangements at these sites.

Figure 3.

Frequency of DNA breakage in normal thyroid cells from RET/PTC+ or PTC− patients. (A) The frequency of blunt-ended double-stranded DNA breaks per 100 cells within RET, FHIT, ALK, 12p12.3 or G6PD regions. (B) The frequency of all DNA breaks per 100 cells includes single-stranded nicks, 3′-overhang, 5′-overhang and blunt-ended double-stranded DNA breaks within RET, FHIT, 12p12.3 or G6PD regions. Both plots are represented by the average of at least three replicated experiments per patient from six RET/PTC+ patient tissues and 7 PTC− tissues. Each diamond represents the average breakage frequency for an individual patient with the horizontal bar representing the mean of all patients in each group. All statistical analyses were performed by two-tailed Student’s t-tests. Asterisk represents P < 0.001 as compared to the respective PTC−.

Next, we chose to focus only on DNA breakage at RET as it is the common partner gene in 19 rearrangements known to cause thyroid cancer (9). By analyzing all types of DNA breakage within RET in each of the normal tissues from the two patient groups, we found that there was no significant difference in the frequency of breakage between these two groups (Figure 3B). There was also no significant difference in total DNA breakage between groups, when analyzing FHIT, 12p12.3 or G6PD. This indicates that blunt-ended DNA breaks are increased specifically in the normal cells of patients with RET/PTC-driven tumors, and suggests that blunt-ended, double-stranded DNA breaks contribute to the formation of RET/PTC rearrangements.

Blunt-ended DNA breaks are the only type of double-stranded DNA breaks increased in normal cells of RET/PTC+ patients compared to RET/PTC− patients

In addition to single-stranded DNA nicks, there are three types of DNA double-stranded break ends: staggered ends with either 5′- or 3′-overhangs and blunt ends. We observed that when comparing the amount of overall DNA breaks at RET intron 11 to that of blunt-ended double-stranded breaks, the vast majority of DNA breaks in each patient tissue were not blunt-ended double-stranded DNA breaks (Figure 3A compared to B). DNA double-stranded breaks must occur for the formation of chromosomal rearrangements, therefore, this prompted us to examine the distribution of DNA breaks among the four types of break ends, and to shed light on the possible substrate required for the formation of RET/PTC rearrangements.

We utilized the activity of the Klenow fragment of E. Coli DNA polymerase I to convert both 5′- and 3′-overhang to blunt-ended double-stranded breaks, and the Klenow fragment mutant (3′–5′ exo-) lacking 3′–5′ exonuclease activity to convert only 5′-overhang ends to blunt ends. The number of DNA breaks detected from treatment of genomic DNA with these enzymes followed by the blunt-end double-stranded detection protocol (Supplementary Figure 1A, available at Carcinogenesis Online, right panel) allows us to detect all four types of DNA breaks. To test the feasibility of our methods, genomic DNA from HTori-3 cells was digested with various restriction enzymes to generate either single-stranded nicks (by Nt.BstNBI digestion), 5′-overhang (BbvI digestion), 3′-overhang (BanII digestion) or blunt-ended (HaeIII digestion) double-stranded DNA breaks. The digested DNA was subjected to the four protocols listed in Supplementary Figure 1A, available at Carcinogenesis Online. In the RET intron 11 region, all three double-stranded breaks were detected as expected in Klenow fragment-treated DNA (Supplementary Figure 3, available at Carcinogenesis Online, lane 2), and only 5′ overhang and blunt-ended breaks were measured in Klenow (3′–5′exo-)-treated samples (lane 3). Lanes 1 and 4 showed the expected products for all four types of breaks and blunt-ended breaks, respectively, when using the all break detection method and the blunt end detection method. All four products were verified by Sanger sequencing to be located at the expected restriction enzyme cut sites in RET intron 11. These results demonstrate the specificity of our assay to detect single-stranded nicks, 3′-overhang, 5′-overhang and blunt-ended double-stranded DNA breaks.

Applying these procedures to measure DNA breaks within RET intron 11 from the normal thyroid cells surrounding RET/PTC-driven tumors and from patients without PTC, no significant difference between the two patient groups was observed in the amount of all three types of double-stranded DNA breaks combined (Figure 4A, left panel). Similarly, the combination of 5′-overhang and blunt-ended double-stranded DNA breaks showed no significant difference between the two groups (Figure 4A, right panel). When individual types of DNA break were evaluated, blunt-ended, double-stranded DNA breaks were the only type of DNA break increased in normal thyroid cells of patients with RET/PTC-driven tumors compared to PTC− patients, and no significant difference was observed when comparing single-strand DNA breaks, 3′-overhang or 5′-overhang DNA breaks between the two patient groups (Figure 4B). This observation suggests that blunt-ended double-stranded DNA breaks could be the preferred substrate for rearrangement formation, and that a test for blunt-ended double-stranded DNA breakage at RET could offer an early indication of susceptibility to PTC.

Figure 4.

Detection of single-stranded, 3′-overhang, 5′-overhang and blunt-ended double-stranded DNA breaks at the RET intron 11 region in normal thyroid cells from RET/PTC+ or PTC− patients. (A) The breakage frequency is represented for all double-stranded DNA breaks or the combination of 5′-overhang and blunt-ended DNA breaks. All double-stranded breaks includes 3′-overhang, 5′-overhang and blunt-ended DNA breaks. (B) The breakage frequency of each DNA break type is shown and represented as DNA breaks/100 cells. Diamonds represent the average DNA breaks from at least three experimental replicates of each the 6 RET/PTC+ patient tissues and 7 PTC− tissues analyzed. The horizontal lines represent the mean of each patient group. Asterisk represents P < 0.001 as compared to PTC−.

Discussion

In this study, we examined whether fragile site breakage could be used to assess the consequence of long-term, low-dose chemical exposure and predict susceptibility to chromosomal rearrangement formation. We found that low-dose, non-cytotoxic exposure to two common environmental chemicals, benzene and DEN, generate significantly more DNA breaks within RET intron 11 compared to untreated cells. We also found that both benzene and DEN induced significantly more blunt-ended double-stranded DNA breaks within RET intron 11 at a rate similar to that of APH treatment, which we have shown can lead to RET/PTC rearrangements (2). Another study demonstrated that H₂O₂ can induce DNA double-stranded breaks and contribute to RET/PTC rearrangements (47), also suggesting the role of chemical exposures on DNA breakage at RET. Therefore, our study demonstrates the importance of investigating these and other low-dose environmental chemicals and their role in the formation of non-radiation associated RET/PTC rearrangements in thyroid cancer. While the HTori-3 cells used for this analysis were immortalized by SV40 and therefore may have altered DNA damage checkpoint and repair, we previously showed (41) that in normal human hematopoetic stem and progenitor CD34⁺ cells, similar, non-cytotoxic levels of benzene, DEN, etoposide and doxorubicin can induce significant DNA breakage at fragile site genes including RET in the presence of optimal DNA repair, indicating the relevance of DNA fragility induced by these chemicals. Further, our results suggest the need for more sensitive tests to screen environmental chemicals, as genes within fragile sites are particularly susceptible to damage.

Secondary cancers following chemotherapy and radiation treatment are a known risk (48) and the ability to better monitor and predict deleterious consequences following treatment of primary cancer is needed. Here, we found that low dose etoposide and doxorubicin, at levels 300–1000-fold lower than those found in the plasma of patients following chemotherapy (43,44), can cause a significant increase in DNA breakage at genes in fragile sites such as RET and FHIT (Figure 2). The low concentrations used in our study mimic the residual conditions in which cells survive chemotherapeutic treatment but are still exposed to these drugs. These results indicate that DNA breakage at genes within fragile sites can occur in surviving cells, with the potential to form cancer-causing gene rearrangements found in secondary cancers. Therefore, the capability of detecting DNA breakage at cancer-specific rearrangement-participating gene regions will be important for patients about to undergo chemotherapy. Identifying those at high risk before administering chemotherapy would help guide treatment decisions toward those that are less likely to affect DNA breakage.

While diagnostic techniques for detecting cancer formation have greatly improved, there is still an urgent need for additional biomarkers to predict cancer susceptibility. Our results demonstrate that normal thyroid cells of patients with RET/PTC-driven tumors have significantly increased blunt-ended, double-stranded DNA breaks within RET, and suggest that this difference could be utilized to predict the propensity to form RET/PTC rearrangements. This is the first study to show significantly higher double-stranded DNA breakage at a cancer driver gene in normal cells of patients with cancer driven by the respective gene. A test to detect DNA breakage at RET could offer an early indication of susceptibility to PTC, and facilitate prompt prevention and timely treatments. Our data suggests that an individual’s susceptibility to fragile site breakage and rearrangement formation is due to the combined effects of both chemical exposures and their susceptibility to these exposures. Therefore, this test could also be used to evaluate individuals at high risk for PTC, either due to environmental exposures and/or unfavorable genotypes. A diagnostic tool to measure individual susceptibility to PTC could improve early detection and possible long-term clinical outcomes. Further, more than half of cancer-specific recurrent chromosomal rearrangements possess DNA breakpoints located within at least one fragile site (49). Therefore, the use of a test for DNA breakage within fragile regions could be expanded to other breakage-sensitive regions in the genome to predict the propensity to form a variety of cancer-specific rearrangements involving these genes. This susceptibility test would be widely helpful to tailor chemotherapy regimens, to monitor high-risk populations and patients in remission, and to improve screening tests for other environmental toxins and stress factors.

Various studies have demonstrated that NHEJ and MMEJ are the DNA repair pathways required to generate chromosomal translocations in human cancers (31,32,34–36). The types of DNA break ends can influence the characteristics of rearrangement fusion points (36). Sequences at the fusion points of RET/PTC rearrangements often appear as short deletions or duplications (37,38,40). The significant amount of blunt-ended double-stranded DNA breaks found in normal cells of patients with RET/PTC rearrangements could represent the immediate substrates for the NHEJ pathway to form rearrangements. Ghezraoui et al. (36) demonstrated the formation of chromosomal translocations following TALEN-induced breakage (with 5′-overhang ends); however, the frequency of these translocations was 3-fold lower than that of Cas9-induced breakage (with blunt ends), which further suggests the important role of blunt-ended double-stranded DNA breaks in the formation of chromosomal translocations. Knockdown of components within the NHEJ pathway decreased the formation of TMPRSS2 fusion transcripts in human prostate cells, which demonstrates NHEJ as the major repair mechanism required to generate fusion genes (33).

Our study provides important insight into the potential for low-dose exposure to environmental chemicals to induce fragile sites and promote the formation of DNA breaks at genes involved in cancer specific chromosomal rearrangements. Further, it provides a foundation for using DNA fragile site breakage to predict susceptibility to chromosomal rearrangements, which would be valuable to monitor high-risk patient groups, tailor chemotherapy regimens, and increase the sensitivity of screening for carcinogenic chemicals. These results also further implicate NHEJ as the major repair mechanism required to generate chromosomal translocations founds in cancers such at PTC. To recapitulate these results in peripheral blood mononuclear cells of patients with RET/PTC-driven tumors compared to healthy individuals would advance these results to the development of a DNA diagnostic test. Paz-Elizur et al. (50) has demonstrated that the activity of DNA repair enzymes such as OGG in PBMCs is a sufficient surrogate for DNA repair activity in lung tissues. As fragile site breakage is the result of the combination of detrimental exposures and insufficient DNA repair processes, this suggests that PBMCs could serve as a sufficient surrogate for the analysis of fragile site breakage in a variety of tissues. To expand on these results and use DNA fragile site breakage as a tool to predict chromosomal rearrangements in a variety of solid tumors would be immensely valuable in advancing the field of clinical oncology diagnostics.

Supplementary material

Supplementary data are available at Carcinogenesis online.

Funding

National Institute of Health (RO1GM101192) and National Cancer Institute (RO1CA113863).

Abbreviations

DEN

diethylnitrosamine

PTC

papillary thyroid carcinoma

NHEJ

non-homologous end joining