GSTT1 Deletion Is Related to Polycyclic Aromatic Hydrocarbons-Induced DNA Damage and Lymphoma Progression

The interrelationship between genetic susceptibility and carcinogenic exposure is important in cancer development. Polymorphisms in detoxification enzymes of the glutathione-S-transferases (GST) family are associated with an increased incidence of lymphoma. Here we investigated the molecular connection of the genetic polymorphism of GSTT1 to the response of lymphocytes to polycyclic aromatic hydrocarbons (PAH). In neoplastic situation, GSTT1 deletions were more frequently observed in lymphoma patients (54.9%) than in normal controls (42.0%, P = 0.009), resulting in an increased risk for lymphoma in individuals with GSTT1-null genotype (Odds ratio = 1.698, 95% confidence interval = 1.145–2.518). GSTT1 gene and protein expression were accordingly decreased in GSTT1-deleting patients, consistent with activated profile of cell cycle regulation genes. Mimicking environmental exposure using long-term repeat culture with low-dose PAH metabolite Hydroquinone, malignant B- and T-lymphocytes presented increased DNA damage, pCHK1/MYC expression and cell proliferation, which were counteracted by ectopic expression of GSTT1. Moreover, GSTT1 expression retarded xenograft tumor formation of Hydroquinone-treated lymphoma cells in nude mice. In non-neoplastic situation, when zebrafish was exposed to PAH Benzo(a)pyrene, molecular silencing of gstt1 enhanced the proliferation of normal lymphocytes and upregulated myca expression. Collectively, these findings suggested that GSTT1 deletion is related to genetic predisposition to lymphoma, particularly interacting with environmental pollutants containing PAH.


Introduction
During the past decades, the incidence of lymphoma has been significantly increased, ranging it among the ten most frequent cancers [1]. The etiologies of lymphoma remain largely undetermined. However, epidemiological studies revealed that exposure to environmental pollutants is a susceptibility factor [2]. Polycyclic aromatic hydrocarbons (PAH) represent the main components of environmental pollutants that have genotoxic and carcinogenic properties.
Genetic polymorphisms in detoxification enzymes are important determinants of individual variation in cancer risk. Glutathione Stransferases (GST) are the major detoxification enzymes in humans. As phase II biotransformation enzymes, GST catalyze the conjugation of reduced glutathione to electrophilic centres on a wide range of substrates, including activated exogenous molecules like PAH.
Several GST polymorphisms commonly occurring in humans are associated with an increased susceptibility to cancers, when combined with environmental factors. Recently, the role of GST genotypes in the pathogenesis of lymphoma has been addressed [3]. GSTT1 is an important member of GST family and involved in the detoxification of various carcinogens, particularly PAH. Evidence of an elevated risk for lymphoma in individuals with GSTT1-null homozygotes has been reported [4,5,6]. Proposed reasons could include an impaired neutralization of reactive oxygen species or reduced deactivation of carcinogenic intermediates of PAH. However, the exact molecular connection between GSTT1 deletions and lymphoma development remained to be investigated.
In the present study, we examined the genetic polymorphisms of GSTT1 in Chinese patients with lymphoma in comparison with a health control cohort, correlating the GSTT1-null genotype with the progression of lymphoma cells and the proliferative behavior of normal lymphocytes under the exposure of PAH. Our results showed that GSTT1 deletion could be a potential risk factor of lymphomagenesis. Genetic susceptibility may interact with the genotoxic effect of environmental carcinogens to eventually predispose to lymphoma.

Ethics Statement
Written informed consent was obtained from all the patients (from the next of kin, caretakers, or guardians on the behalf of the minors/children patients) in accordance with the Declaration of Helsinki. The study was approved by the Shanghai Rui Jin Hospital Review Boards. Animals were used according to the protocols approved by the Shanghai Rui Jin Hospital Animal Care and Use Committee.

Patients
Two hundred and four patients with lymphoma [103 diffuse large B-cell lymphoma (DLBCL) and 101 T-cell lymphoma (TCL) cases], 127 men and 77 women aged 14 to 82 years were included in this study. Histological diagnoses were established according to the World Health Organization classifications. Frozen tumor specimen was available in 114 lymphoma patients and 40 age-and sex-matched reactive hyperplasia cases.

Cell Lines and Reagents
The B-lymphoma cell line Namalwa and T-lymphoma cell line Jurkat were obtained from American Type Culture Collection. Hydroquinone (Sigma-Aldrich) was dissolved in normal saline before use. Benzo(a)pyrene (BaP, Sigma-Aldrich) was dissolved in dimethyl sulfoxide (DMSO) as a stock solution of 400 mg/ml.

Cell Proliferation and Flow Cytometric Assay
Cell growth was measured by MTT assay and cell proliferation was determined by 5-ethynyl-29-deoxyuridine (EdU) incorporation assay using Cell-Light EdU kit (Ribobio Co., Ltd., China) at 48 h. Cell cycle and cell apoptosis were assessed at 48 h as previous reported [7].

Genome-wide Copy Number Variation (CNV) Analysis
Genomic DNA was extracted using Wizard Genomic DNA Purification Kit (Promega). Genome-wide CNV genotyping was performed on frozen tumor samples of 25 DLBCL, 20 TCL and 8 reactive hyperplasia cases, using Human 610-Quad_v1 (610 k SNP probes) or 660 W-Quad_v1 (660 k SNP probes) DNA Analysis BeadChips. Regions were determined based on the Log R Ratio (LRR) of the signal intensity and B allele frequency (BAF) of genotyping call from the sample using platform of GenomeStudio V2011.1 with CnvPartition 3.1.6 (Illumina). All the data is available on NCBI (Accession number GSE47357).

GSTT1 Genotyping and Expression
GSTT1 deletion was detected on frozen tumor (114 cases) and peripheral blood (the rest 90 cases without frozen tumor specimen) of lymphoma patients by multiplex polymerase chain reaction (PCR) method, using albumin gene as an internal positive control, as previously reported [8]. The normal control group comprised 205 unrelated healthy volunteers. Blood samples were collected and leukocytes were isolated after hypotonic lysis of erythrocytes. Genomic DNA was extracted and amplified using the primers: GSTT1:59-TTCCTTACTGGTCCTCACATCTC-39 and 59-TCACCGGATCATGGCCAGCA-39, and albumin:59-GCCCTCTGCTAACAAGTCCTAC-39 and 59-GCCCTAAAAAGAAAATCGCCAATC-39.
Total RNA was extracted using Trizol reagent and reversetranscribed using PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa). Real-time PCR was performed on frozen samples of lymphoma and reactive hyperplasia patients, using ABI PRISM 7900HT and specific probes for GSTT1 (Assay ID: Hs01091675_g1) and GAPDH (Life Technologies). A relative quantification was calculated using the 2 2DDCT method.

Gene Network and Pathway Analysis
Human Genome U133 Plus 2.0 Array GeneChip microarray (Affymetrix) was performed on tumor samples of 8 DLBCL patients and analyzed by Expression Console software (Partek GS 6.5, Affymetrix). The data is available on NCBI (Accession number GES47355). Human LncRNA Microarray V2.0 (Arraystar Inc.) was performed on tumor samples of 8 TCL cases and analyzed by Agilent Feature Extraction Software (Agilent Technologies).
Genes were subsequently filtered by comparing their expression levels between the GSTT1-deleting and non-deleting patients. Statistical differences were calculated and the genes with P,0.05 were analyzed for enrichment of KEGG pathways using Database for Annotation, Visualization and Integrated Discovery (DAVID v6.7, http://david.abcc.ncifcrf.gov) for network composition analyses. Genes of pathway(s) significantly involved in both DLBCL and TCL were hierarchical clustered using MeV v4.8.1 (Dana-Farber Cancer Institute).

Comet Assay
DNA damage was determined by the comet assay with Reagent Kit for Single Cell Gel Electrophoresis Assay (Trevigen, Inc.). In addition to frank DNA strand breaks, oxidised bases were measured by conversion to breaks using endonuclease III (recognizing oxidised pyrimidines) or formamidopyrimidine DNA glycosylase (FPG, specific for oxidised purines). Measurements of comet parameters % DNA in the tail, tail length and tail moment were obtained. Net enzyme-sensitive sites were calculated by subtracting the comet score after incubation with buffer alone from the score with enzyme.

Tissue Array
A human lymphoma tissue array (NHL482) was obtained from US Biomax, Inc. GSTT1 and p53-binding protein 53 BP1 expression were scored semi-quantitatively based on staining intensity and distribution using the immunoreactive score, as previously reported [7].

Immunohistochemistry and Immunofluorescence Assay
Immunohistochemical analyses were carried out on 5-mmparaffin sections or acetone-fixed cells with an indirect immunoperoxidase method using the antibodies against GSTT1, 53 BP1 (Abcam), Ki67 (Dako) and MYC (Abcam). Immunofluorescence assay was performed on acetone-fixed cells using mouse antihuman-cH2AX antibody followed by donkey anti-mouse-IgG antibody, as well as rabbit anti-human-53 BP1 and rabbit antihuman-pCHK1 antibody followed by donkey anti-rabbit-IgG antibody (Abcam). Nuclei were counterstained with DAPI.

Western Blot
Western blot was performed as previously described [7]. Actin (Sigma) and LaminB (Abcam) were used to ensure equivalent loading of total and nuclear protein, respectively. Antibodies against GSTT1 and CHK1 were obtained from EPITOMICS. Anti-pCHK1, pCHK2 and CHK2 antibodies were from Cell Signaling. Anti-MYC and 53 BP1 antibodies were from Abcam. Horseradish peroxidase-conjugated goat anti-mouse-IgG and goat anti-rabbit-IgG were from Santa Cruz Biotechnology Inc.

Tumorigenicity Assay in Murine Models
Five-week-old female BALB/c nude mice were obtained from Shanghai Laboratory Animal Center. Mice were injected subcutaneously into the right flank with lymphoma cells. For each cell line, mice were divided into 3 subgroups. Namalwa cells were injected with 1610 7 , 2610 6 and 5610 5 , and Jurkat cells were injected with 4610 7 , 1610 7 and 2610 6 , respectively. The number of the tumors formed was determined until 4 weeks after injection. Then mice were sacrificed, with tumor tissue samples fixed in formaldehyde and further processed for paraffin embedding.

Cloning and Plasmid Construction in Zebrafish
Adult zebrafish (Danio rerio) were maintained following established guidelines [9] at 28uC on a 14 h:10 h light:dark cycle. Zebrafish gstt1a and gstt1b genes were identified based on homology to human GSTT1. The specific primers were designed according to genomic sequence in the UCSC data base (University of California, Santa Cruz) to amplify part of gstt1a and gstt1b genes: myca:59-CGGGATCCTTTCCAAGAACTCCCACCCC-39 and 59-CCCTCGAGTTTAATCTAGGGCTGCGCAG-39. The amplified PCR product was purified and subcloned into pCS2 + for subsequent in vitro synthesis of mRNA.
The vectors pCS2 + -MO-gstt1a and pCS2 + -MO-gstt1b were linearized with XhoI and transcribed in vitro with SP6 RNA polymerase in the presence of m7G (59)ppp(59)G (Ambion) to produce capped transcripts and were microinjected into one-cell stage embryos for MO efficiency evaluation.

Transmission Electron Microscopy
The embryos of zebrafish were fixed overnight with 2% glutaraldehyde/0.1 M phosphate-buffered saline (pH 7.3) at 4uC. The embryos were then post-fixed in 1% osmium tetroxide at 4uC for 1 h, rinsed thoroughly with distilled water, dehydrated by graded ethanol and freeze-dried. The samples were sputtercoated in Epon812 (TAAB Laboratories) and ultrathin sections were prepared, stained with uranyl acetate and lead citrate, and examined with a PhilipsCM120 transmission electron microscopy (Philips).

Statistic Analysis
Assays were set up in triplicate and the results were presented as Mean6S.E. Variance between the experimental groups was determined by two-tailed t-test or Mann-Whitney test. The logistic regression test was used to compare the differences in genotype frequencies between patients and controls adjusted by sex and age. Differences were considered significant when the 2-sided P,0.05.

GSTT1-null Genotype is Frequently Observed in Lymphoma Patients
A homozygous loss in chr22q11.23 spanning GSTT1 gene (22,706,139 bp to 22,715,284 bp, NCBI Build 36) was detected in 15/25 (60%) of DLBCL and 7/20 (35%) of TCL patients, comparing with 1/8 (12.5%) of reactive hyperplasia cases ( Figure 1A). Figure 1B depicted data for the region of 22,500,000 bp to 22,900,000 bp of chromosome 22. In the normal control, the LRR was distributed around zero corresponding to DNA copy number 2, whilst the BAFs were clustered around values of 0, 0.5 and 1 that correspond to the diploid genotypes AA, AB and BB. The GSTT1-null genotype presented a much more complex scenario with extensive genomic rearrangements leading to considerable variation in the SNP data.
Gene expression profile was assessed by microarray in frozen tissue sample of 8 DLBCL and 8 TCL patients with the history of PAH exposure (four each for GSTT1-deleting and non-  To further elucidate its biological function in lymphoma, GSTT1 gene was transfected to GSTT1-negative Namalwa and Jurkat cells. Comparing with the negative control (FU), lymphoma cells expressing GSTT1 (GSTT1) showed increased levels of gene and protein expression, as revealed by semi-quantitative PCR ( Figure 3A) and by immunostaining assay (Figure 3B), respectively.
Hydroquinone, common metabolites of PAH [11], was used to treat lymphoma cells. Fifty percent of growth inhibition (IC50) was first measured in both cell lines at concentrations ranging from 1.25 to 40 mM. The IC50 value was 14.4 mM in Namalwa cells and 18.7 mM in Jurkat cells, respectively. To mimic the environmental exposure status, lymphoma cells were then cultured with low-dose Hydroquinone (at a concentration of approximately 10% of IC50 with minimal cytotoxicity, 1 mM for Namalwa GSTT1/FU cells and 1.5 mM for Jurkat GSTT1/FU cells, respectively) in a repeat manner [12]. Briefly, cells were treated once a week with Hydroquinone for 24 h and weekly for 4 times in total. Normal saline was referred as the solvent control.
DNA strand breaks were assessed using comet assay, and in addition DNA oxidation damage, namely oxidised pyrimidines and purines, using lesion specific enzymes Endo III and FPG. As manifested by increased parameters % DNA in the tail and prolonged tail length, DNA damage was less intensive in Hydroquinone-treated GSTT1-expressing lymphoma cells (GSTT1) than in the negative control cells (FU). Tail moment was further assessed by parameters % DNA6tail length. The results showed that tail moment, as well as EndoIII-and FPGsensitive sites, were significantly reduced in the GSTT1 groups than in the FU groups (Namalwa, P = 0.037, P = 0.042 and P = 0.042, Jurkat, P = 0.041, P = 0.049 and P = 0.049, respectively, Figure 3C).
cH2AX and 53 BP1 are sensitive markers of DNA damage [13]. As detected by immunofluorescence assay, FU cells presented increased nuclear levels of cH2AX and 53 BP1, which were significantly prohibited in GSTT1 cells ( Figure 3D). These data indicated that DNA damage is constitutively activated in response to PAH and could be protected by GSTT1 expression.

GSTT1 Inhibits PAH-mediated Lymphoma Cell Proliferation
Upon treatment with normal saline, there is no difference of cell growth between the FU and GSTT1 groups. However, cell growth was enhanced in Hydroquinone-treated FU cells, which could be significantly reduced by ectopic expression of GSTT1 (Namalwa, P = 0.045 and Jurkat, P = 0.043, respectively, Figure 4A). Cell proliferation was further determined by EdU assay. Comparing with the FU groups, EdU-positive cells in the GSTT1 groups were accordingly reduced (Namalwa, P = 0.043 and Jurkat, P = 0.039, respectively, Figure 4B). Cell cycle analysis showed a lower percentage of S-phase cells in the GSTT1 groups than in the FU groups (Namalwa, 17.560.52% vs 14.360.55%, P = 0.013, and Jurkat, 14.160.44% vs 11.860.54%, P = 0.032, respectively). As for cell apoptosis, neither Namalwa nor Jurkat cells showed obvious change in the percentage of ANX-V-positive cells between the FU and GSTT1 groups when treated with Hydroquinone ( Figure 4C).
By Western blot, expressions of cell cycle regulation proteins were assessed with or without Hydroquinone treatment. In Hydroquinone-treated Namalwa and Jurkat cells, GSTT1 expression was consistent with downregulation of MYC, pCHK1 and nuclear 53 BP1, while pCHK2, total CHK1 and CHK2 levels remained constant ( Figure 4D).

GSTT1 Retards Tumor Formation of PAH-treated Lymphoma Cells
To assess the carcinogenic potential in vivo, the FU and GSTT1 cells were treated with Hydroquinone as mentioned above and injected subcutaneously in nude mice ( Figure 5A). Expression of GSTT1 could significantly prolong the latency of tumor formation. Of note, 2610 6 Namalwa cells derived from the FU group could form tumors in 10 of 12 nude mice in 4 weeks. However, when 2610 6 cells from GSTT1 group were injected, only 4 tumors were found in 12 nude mice (P = 0.013). Similar results were obtained in Jurkat cells: tumor were formed in all mice at 4 weeks injected with 1610 7 FU cells, but only 6 of 12 mice in those treated with the same amount of GSTT1 cells (P = 0.014).
To search for in situ evidence of DNA damage and tumor cell proliferation, immunoflurescence assay of cH2AX, 53 BP1 and pCHK1, as well as immunohistochemistry assay of Ki67 and MYC were performed on mice tumor sections. In parallel with in vitro results, all markers were increased in the FU groups, compared with the GSTT1 groups following Hydroquinone treatment ( Figure 5B).

GSTT1 Knock-down Stimulates Lymphocyte Proliferation in PAH-treated Zebrafish
After searching the zebrafish genome data base (Zv9), two GSTT1 genes were identified, referred as gstt1a and gstt1b, sharing part of nucleotide acids identical to human GSTT1 gene. To confirm the evolutionary conservation of GSTT1, we investigated the distribution of genes located adjacent to the GSTT1 locus in zebrafish and in human genomes. As shown in Figure S1A, 9 genes, along with GSTT1, define an approximately 840-kb genomic region on human chromosome 22, which is syntenic to the zebrafish gstt1a genomic locus on linkage group 8 and zebrafish gstt1b on linkage group 21. Thus, zebrafish gstt1a and gstt1b are evolutionarily conserved orthologs of human GSTT1.
Then the temporal and spatial expression patterns of gstt1a and gstt1b were examined in zebrafish embryos from 0.25 h to 120 hpf by WISH using digoxigenin-labeled antisense RNA probe. Strong signals were detected in the blastomeres of the two-cell stage and the shield period (6 h, Figure S1B), indicating that both gstt1a and gstt1b transcripts were maternally derived and played a role in early embryonic development. Although both genes exhibited similar patterns of expression during the early stage, differential expression was observed after 30 hpf: high levels of gstt1a transcripts were mostly restricted to the liver (2d), while gstt1b transcripts gradually disappeared after 30 hpf and was no longer observed after 3d ( Figure S1B).
To further determine the function of GSTT1 on normal lymphocytes, morpholino was applied to efficiently block gstt1a and gstt1b gene expression in zebrafish. Injection of MO of gstt1a at the dosage of 8 ng per embryo, but not mismatch oligo control, was able to completely suppress the EGFP expression of coinjected capped gstt1a-EGFP mRNAs reporter ( Figure S1C, left panels), indicating that the antisense RNA that could effectively silence gene expression. Similar results were obtained in gstt1b ( Figure S1C, right panels).
Exposure of BaP, a carcinogenic PAH, was conducted at the concentration of 10 mg/L, based on previous experiments investigating the teratogenicity of PAH on zebrafish [14,15]. Zebrafish embryos were microinjected with gstt1a and gstt1b morpholinos or 5 bp mismatch oligo controls, and treated with  BaP from 24 hpf to 4 d. DMSO was used as the solvent control at a final concentration of 0.1%. At least 100 embryos were included in each group.
Comparing with those injected with mismatch oligos, gstt1a and gstt1b silenced embryos showed an increase of the rag1 signal in the thymus under exposure to BaP, as detected by WISH ( Figure 6A, right panels). No significant difference of the rag1 signal was observed in embryos treated with DMSO ( Figure 6A, left panels).
To confirm that MYC activation is also involved in PAHassociated stimulation of lymphocyte proliferation in zebrafsh, the expression of myca, zebrafish homologue of human MYC, was assessed in BaP-treated embryos at 5 d. In accordance with rag1 expression, myca expression was significantly higher in embryo injected with gstt1a and gstt1b morpholinos than those injected with mismatched oligos by WISH ( Figure 6B, left panels) and by semiquantitative PCR ( Figure 6B, right panels).
The structure of the thymus in 5 d-embryos was examined by transmission electron microscopy ( Figure 6C). Ultrastructural analysis of both cells (microinjected with gstt1a and gstt1b morpholino or oligo control) revealed no sign of apoptosis. At a higher magnification, abundant euchromatin and robust mitochondrias were observed in BaP-treat gstt1-silenced lymphocytes, possibly related to a hyper-proliferative status of the cells.
Therefore, in addition to lymphoma cells, inactivation of GSTT1 was able to confer a proliferative advantage induced by PAH in their normal counterparts.

Discussion
Although environmental factors are proven relevant for tumorogenesis, few genetic variations with confirmed associations to date and the difficulty in accurately assessing exposures are main challenges to evaluate the interaction between gene and environment in cancer [16]. Our study provided evidence that genetic polymorphisms in GSTT1 gene, both in neoplastic and non-neoplastic situation, could modulate the response of the lymphocytes to the major component of environmental pollutants PAH and might link to lymphoma development.
The GSTT1-null genotype was more prevalent in lymphoma patients than in normal controls, conferring approximately a 1.7fold increase in the risk of lymphoma. This coincides with epidemiological studies from Western [4], African [5], and other Asian countries [6], although the incidence of GSTT1 deletion varied among the geographical areas. The genetic polymorphism was associated with loss of gene and protein expression, indicating that GSTT1 is functionally impaired in lymphoma.
It is previously reported that deletions in the GSTT1 gene contribute to individual susceptibility to PAH-induced DNA damage and carcinogenesis [17,18]. Our study further showed in lymphoma that introduction of GSTT1 to GSTT1-negative tumor cells is associated with increased DNA stability and repair of oxidative DNA damage in response to PAH. These circumstances fit to the idea that GSTT1 was involved in the susceptibility of individuals to PAH-induced DNA damage, loss of which favoring the accumulation of cytogenetic aberrations and thereby leading to the initiation of lymphoma. Therefore, genetic polymorphisms in detoxification enzymes might account for individual variation in lymphoma risk and should be considered in a more complex scenario involving the gene-environment interactions.
GSTT1 can modulate multiple cellular processes, including cell proliferation and cell death [19]. Among the lymphoma patients with the history of PAH exposure, instead of cell apoptosis, the GSTT1-deleting cases displayed a genomic profile of cell cycle progression, referring dysregulation of cell proliferation as the major target of GSTT1 deletions in lymphoma. Experimentally, in GSTT1-negative lymphoma cells, expression of GSTT1 significantly prohibited PAH to enhance tumor cell growth, with tumor aggressiveness accordingly decreased in murine models. These data thus suggested a potential role for GSTT1 in protecting against lymphoma cell proliferation provoked by PAH.
MYC is essential for cell proliferation [20] and is a protooncogene frequently upregulated in lymphoma [21]. More importantly, MYC itself can localize onto sites of active DNA replication and directly controls S-phase cell progression [22]. MYC was activated following PAH treatment in our study, particularly in GSTT1-negative lymphoma cells, corresponding to increased S-phase cells, enhanced cell proliferation and in vivo tumorigenicity, indicative the possible involvement of MYC on PAH-associated cell cycle dynamics and lymphoma progression in GSTT1-null status.
The MYC-induced DNA damage response acts as a doubleedged sword in tumor progression. CHK1 and CHK2 are key factors involved in the replication stress response and controlled by MYC [23]. Indeed, activation of CHK1 is essential for tumor maintenance, while CHK2 activity constitutes a barrier to malignant transformation. As previously reported in Em-myc lymphoma models, tumor cells present increased levels of CHK1 phosphorylation, in turn limits MYC-induced apoptosis. In the clinical setting, lymphoma patients exhibit a striking correlation between high levels of MYC and CHK1 [24]. This was also proven by us in PAH-treated lymphoma cells, where MYC might ensure proliferative advantage through selectively activating CHK1. Recent reports have identified MYC-positive lymphoma as a subtype with poor disease prognosis [25], even resistant to high-dose chemotherapy [26] and newly developed bio-therapeutic agent [27]. Since MYC is difficult to be targeted directly, CHK1 inhibitors could thus become attractive candidates for therapeutic intervention on MYC-driven malignancies and warrant further investigation.
Genetic factors that impair DNA repair can increase the likelihood of pre-neoplastic changes [28]. This is particularly obvious when environmental factors were existed, as a previous report showing that the t(14;18)-positive clones are prominent in individuals exposed to pesticides and correlated with a higher risk of t(14;18) lymphoma [29]. In addition to lymphoma cells and murine xenograft models, we used zebrafish as an animal model to verify the cooperative effect of the genetic and environmental factor on their normal counterparts. In GSTT1-knock-down zebrafish, although may not initially be lethal, genomic lesions in lymphocytes could be modulated by PAH that promote lymphocyte proliferation and MYC upregulation, which could eventually link to malignant transformation of lymphoma.

Supporting Information
Figure S1 GSTT1 is evolutionarily conserved and expressed ubiquitously during embryonic development. A: Comparison of the syntenic relationship of the zebrafish gstt1 genes with the human orthologue. Orthologous gstt1 genes symbols were in bold. Other pairs of duplicated genes (e.g. mmp11a and mmp11b) on zebrafish and the human (e.g. MMP11) orthologue are Figure 6. Knock-down of gstt1a and gstt1b promotes lymphocyte proliferation exposed to BaP. A; WISH images showed the rag1 expression in the thymus (arrows) of differently treated 5 dpf embryos. B: In situ analysis of myca at 5 dpf. The morphants showed increased expression of myca in microinjected gstt1a and gstt1b morpholino exposed to BaP (Left panels), semi-quantitative PCR showed similar expression pattern in embyos (Right panels). C: Ultrastructure of thymic lymphocytes from 5 dpf larvae exposed to BaP. Images represent results from three independent experiments and each group contains 30 morphants. doi:10.1371/journal.pone.0089302.g006 underlined. Hs_Chr, Homo sapiens chromosome, Dr_LG, Danio rerio linkage group, Mb, megabase. B: Expression of zebrafish gstt1a and gstt1b in wild-type AB strain embryo during embryonic development. C: Efficiency validation of EGFP reporter expression by gstt1a morpholino and gstt1b morpholino. Images represent the typical outcome of three independent experiments and each group contains 30 morphants. (TIF)

Author Contributions
Conceived and designed the experiments: W-LZ S-JC. Performed the experiments: FY JX. Analyzed the data: Z-HG. Contributed reagents/ materials/analysis tools: X-EJ J-YS YZ J-ML. Wrote the paper: W-LZ FY.