PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Cancer Res. Author manuscript; available in PMC Jul 3, 2013.
Published in final edited form as:
PMCID: PMC3700628
NIHMSID: NIHMS141347
A modified Sleeping Beauty transposon system that can be used to model a wide variety of human cancers in mice
Adam J. Dupuy,1,7 Laura M. Rogers,1 Jinsil Kim,1 Kishore Nannapaneni,2 Timothy K. Starr,3 Pentao Liu,4,7 David A. Largaespada,3 Todd E. Scheetz,2,5 Nancy A. Jenkins,6,7 and Neal G. Copeland6,7
1Department of Anatomy & Cell Biology, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, USA
2Center for Bioinformatics; Computational Biology and Biomedical Engineering, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA 52242
3The Department of Genetics, Cell Biology and Development, Masonic Cancer Center, University of Minnesota, Minneapolis, Minnesota 55455, USA
4Wellcome Trust Sanger Institute, Hinxton, Cambridge, CB10 1SA, UK
5Ophthalmology and Visual Sciences, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA 52242
6Institute of Molecular and Cell Biology, 61 Biopolis Drive, Proteos, Singapore 138673
7Former address: Mouse Cancer Genetics Program, Center for Cancer Research, National Cancer Institute, Frederick, MD 21702, USA
Recent advances in cancer therapeutics stress the need for a better understanding of the molecular mechanisms driving tumor formation. This can be accomplished by obtaining a more complete description of the genes that contribute to cancer. We previously described an approach using the Sleeping Beauty transposon system to model hematopoietic malignancies in mice. Here we describe modifications of the SB system that provide additional flexibility in generating mouse models of cancer. First, we describe a Cre-inducible SBase allele, RosaSBaseLsL, that allows restriction of transposon mutagenesis to a specific tissue of interest. This allele was used to generate a model of germinal center B-cell lymphoma by activating SBase expression with a Aid-Cre allele. In a second approach, a novel transposon was generated, T2/Onc3, in which the CMV enhancer/β-actin (CAG) promoter drives oncogene expression. When combined with ubiquitous SBase expression, the T2/Onc3 transposon produced nearly 200 independent tumors of over 20 different types in a cohort of 62 mice. Analysis of transposon insertion sites identified novel candidate genes, including Zmiz1 and Rian, involved in squamous cell carcinoma and hepatocellular carcinoma, respectively. These novel alleles provide additional tools for the SB system and provide some insight into how this mutagenesis system can be manipulated to model cancer in mice.
The relationship between somatic mutation and cancer is well established. We understand that the step-wise acquisition of mutations over time drives cancer progression. These ideas are well defined and provide a basic framework for the field of cancer genetics, but our ability to precisely identify the gene mutations that contribute to an individual tumor remains limited.
Recent efforts to re-sequence coding exons in human tumors have identified somatic mutations in breast and colorectal cancer (1), pancreatic cancer (2), glioblastoma multiforme (3) and lung cancer (4). These studies found an average between 47 and 80 coding region somatic mutations per tumor. One major implication of this work is that the number of “passenger mutations” greatly exceeds the number of “driver mutations.” Consequently, more sequencing will be required to gain a comprehensive view of the human cancer genome. This is especially true given the relatively low mutation rate of most cancer genes within a patient population (1).
We recently described a novel transposon-based insertional mutagenesis system, called Sleeping Beauty, which can be used as a cancer gene discovery tool in mice (5, 6). Our initial version of the Sleeping Beauty (SB) system consisted of the RosaSBase knock-in allele (transposase source) and the T2/Onc2 allele (transposon source). A significant number of mice that inherit both alleles die during embryogenesis, and the remaining mice develop aggressive tumors [(for review see (7, 8)]. Mutations in SB-induced tumors are caused and tagged by transposon insertions, facilitating the identification of cancer genes.
The initial SB system had several significant limitations. First, the high rate of embryonic lethality limited production of RosaSBase;T2/Onc2 double-transgenic mice. Second, the initial SB system produced hematopoietic malignancies, yet failed to produce carcinomas, which occur more frequently in the human population (6). These features limited the ability of the SB system to model cancer in mice.
Here we describe a Cre-inducible SB transposase allele, in addition to a novel mutagenic transposon, T2/Onc3. These new components provide additional flexibility to the system, greatly broadening the tumor spectrum that can be produced by in vivo transposon mutagenesis.
Cre-inducible Sleeping Beauty mutagenesis
By restricting transposase expression to a specific tissue, we hoped to reduce or eliminate embryonic lethality and generate tumors in any desired tissue concomitantly. We elected to use a lox-stop-lox approach, in which the RosaSBase allele is under the control of a floxed transcriptional stop cassette (Fig. 1A); an approach already used to produce many lox-stop-lox ROSA26 alleles which express a reporter in a Cre-dependent manner (9, 10).
Figure 1
Figure 1
Generation of the RosaSBaseLsL allele
We used a similar strategy to introduce a floxed stop cassette, consisting of an EGFP cDNA followed by three tandem polyadenylation signals, upstream of the SBase cDNA in the RosaSBase allele, thereby producing the RosaSBaseLsL allele (Fig. 1A). This design should ensure that only EGFP is expressed prior to the removal of the lox-stop-lox cassette by Cre recombination. The RosaSBaseLsL cassette was targeted to the Gt(ROSA)26Sor locus in mouse ES cells, and Southern blotting of ES cell clones confirmed the structure of the RosaSBaseLsL allele (data not shown). Chimeric mice generated by blastocyst injection were bred to C57BL/6J mice to propagate the line.
Next, we tested the function of the RosaSBaseLsL allele. The lox-stop-lox cassette was excised in the germline by crossing RosaSBaseLsL/+ mice to β-actin-Cre transgenic mice. This produced the RosaSBase1loxP allele, which differs from the RosaSBase allele in that a single loxP site lies between the splice acceptor and the SBase cDNA (Fig. 1A). We predicted that the RosaSBase1loxP allele would function in a similar fashion to RosaSBase. Thus, we expected a significant number of RosaSBase1loxP;T2/Onc2 double-transgenic embryos would die during gestation, and that the remaining mice would develop aggressive hematopoietic tumors with a short latency.
To test this hypothesis, we generated a small cohort of animals (n=24) by crossing heterozygous RosaSBase1loxP mice to TG6070 mice homozygous for a T2/Onc2 transposon concatamer array that contains ~200 copies of the transposon (6). Despite our prediction, we did not see evidence of embryonic lethality in this cohort (Supplementary Fig. S1). It is possible that the loxP site and additional sequences in RosaSBase1loxP interfere with the translation of the transcript. However, SBase protein levels in tissues from RosaSBase and RosaSBase1loxP mice were not dramatically different (Supplementary Fig. S1). It should also be noted that RosaSBase;TG6070 double transgenic mice display the least amount of embryonic lethality of the three T2/Onc2 strains previously described (6). Thus, a slight change in the SBase expression level could reduce or eliminate the lethality.
Despite the apparent difference between the RosaSBase and RosaSBase1loxP alleles, all RosaSBase1loxP;TG6070 double-transgenic mice developed hematopoietic tumors (Fig. 1B). Tumor latency (avg. ~59 days) is consistent with the results obtained using RosaSBase as the transposase source (6). Therefore, reduction in transposition efficiency induced by the RosaSBase1loxP allele does not significantly impact tumor latency when combined with the TG6070 T2/Onc2 transposon donor transgene.
Finally, RosaSBaseLsL allele tissue specificity was tested by generating triple-transgenic mice that were heterozygous for RosaSBaseLsL, the T2/Onc2 transposon array and an Aid-Cre knock-in allele. The Aid-Cre allele has a Cre recombinase cDNA fused to the initiation codon of the activation-induced cytidine deaminase gene (Aicda), and should drive Cre-mediated recombination in germinal center B-cells. Triple-transgenic mice and double-transgenic littermate controls that did not inherit the Aid-Cre allele were aged (Fig. 1B). A log-rank test showed that triple-transgenic mice had significantly decreased survival compared to their double-transgenic littermates (p=0.0322) (Fig. 1B). The pathology of moribund mice indicated that over 90% of triple-transgenic mice developed tumors with an average latency of 46 weeks. In contrast, double-transgenic mice were not tumor prone, though many were euthanized due to complicating pathologies such as urinary tract infection (Supplementary Table S1). Similar pathology was not noted in other RosaSBaseLsL;T2/Onc2 cohorts we generated (data not shown) or in another experiment using the RosaSBaseLsL allele (11). Thus, the complicating pathologies seen in double-transgenic control mice are not likely induced by transposition.
The majority of tumors developed by triple-transgenic mice were of B-cell origin, including diffuse large cell lymphomas and follicular lymphomas (Fig. 1C, Supplementary Table S1). RosaSBase1loxP;T2/Onc2 double-transgenic mice predominantly developed T-cell lymphomas, consistent with our previous findings (Fig. 1C) (6). Notably, some of the triple-transgenic mice developed malignancies in other cell types (Fig. 1C).
Carcinomas induced by T2/Onc3 mutagenesis
We previously demonstrated that ubiquitous transposition induces hematopoietic tumors in mice, though transposition could be detected in other cell types (6). While many factors can contribute to bias in tumor spectrum, one that can be easily manipulated is transposon design. The T2/Onc2 transposon contains the 5’ long terminal repeat (LTR) from the murine stem cell virus (MSCV) to drive overexpression of oncogenes (6). The MSCV is derived from a cloned Moloney murine leukemia virus (12) and is likely to drive oncogene expression in hematopoietic cells at a higher rate than in other cell types. We reasoned that swapping the MSCV LTR promoter for one that expresses strongly in epithelial cells should alter the tumor spectrum produced.
We generated a new mutagenic transposon, T2/Onc3, whose structure is identical to that of T2/Onc2 (6) except that the MSCV LTR is replaced with the CAG promoter (Fig. 2A). This promoter consists of the cytomegalovirus immediate early enhancer fused to the chicken ®-actin promoter, and expresses strongly in a variety of cell types in vivo, including epithelial cells (13, 14). We then tested the effect of this change on the tumor spectrum.
Figure 2
Figure 2
Structure and function of the T2/Onc3 transposon
The T2/Onc3-containing plasmid was linearized and injected into C57BL6/J×C3H F1 one-cell embryos. Founder mice were initially identified by Southern blotting. Estimating transposon copy number is important, as this parameter has been shown to have an impact on tumor latency (7). We were previously able to generate high-copy transposon donors (150–300 copies) using the T2/Onc2 transposon (6). However, we were unable to produce high-copy T2/Onc3 transgene founders, despite screening over 340 founder animals (data not shown). We generated two T2/Onc3 transgenic lines from two founder animals (TG12740 and TG12775) by crossing them to wild-type C57BL6/J mice. Using quantitative PCR, we estimate the TG12740 and TG12775 transgenes contain roughly 11 and 28 copies, respectively (Supplementary Fig. S2).
Heterozygous T2/Onc3 transgenic mice from each line were then crossed to homozygous RosaSBase animals to induce ubiquitous transposition. Double-transgenic mice were obtained from these crosses at the expected Mendelian ratio, indicating that ubiquitous transposition of the T2/Onc3 transposon does not induce embryonic lethality (Supplementary Table S2). This was expected, since it was previously shown that the combination of RosaSBase and a low copy transposon donor does not cause embryonic lethality (7).
We generated and aged a cohort of 73 double-transgenic mice, along with 45 RosaSBase single-transgenic littermate controls. Compared to single-transgenic RosaSBase mice, double-transgenic mice had a decreased rate of survival (log-rank, p<0.0001) (Fig. 2B). Histological analysis was performed on all double-transgenic mice for which material was available (n=62) and on 19 randomly selected RosaSBase control mice (Table 1). In contrast to control mice, double-transgenic mice develop a variety of malignancies, with each animal developing 3 independent primary tumors on average. Only 11% of T2/Onc3-induced tumors were lymphomas, compared to ~90% of T2/Onc2-induced tumors (6). Instead, double-transgenic mice developed a wide array of carcinomas, including squamous cell carcinoma of the skin and hepatocellular carcinoma (Table 1). Notably, a significant gender bias was associated with both hepatocellular carcinoma (p=0.0052) and liver adenoma (p=0.00028); male double-transgenic mice were significantly more susceptible to liver malignancies. This is similar to the gender bias seen in human hepatocellular carcinoma (15).
Table 1
Table 1
Pathology of T2/Onc3-induced tumors
Histological examination found 4 cases of metastasis in RosaSBase;T2/Onc3 double transgenic mice (Table 1). Metastasis was associated with a variety of tumor types: melanoma, hepatocellular carcinoma, hemangiosarcoma and salivary gland carcinoma (Fig. 3, Supplementary Fig. S4). In the metastatic hepatocellular carcinoma, numerous macrometastases were detectable in the lungs (Fig. 3B). We generated transposon integration profiles for both the primary liver tumor and one of the large metastatic lesions from the lung using the ligation-mediated PCR strategy subsequently described (Supplementary Fig. S3). Two identical transposon integration events were identified in both tumors, suggesting these tumors share a common origin (Supplementary Table S3). However, the majority of integration events identified in each tumor were unique events.
Figure 3
Figure 3
Metastasis in T2/Onc3-induced carcinoma
Identification of transposon-induced mutations
We utilized a modified ligation-mediated PCR (LM-PCR) method to produce barcoded PCR products from SB-induced tumors that can be directly sequenced on the GS FLX System (11, 16). Briefly, an “amplicon sequencing” approach was used in which unique tags were added to the 5’ ends of the primers used during the final amplification step of the LM-PCR. Additionally, we added a 10-basepair barcode (17) to the 5’ end of the transposon-specific primer, such that it would be sequenced just before the transposon tag (Supplementary Fig. S3). Therefore, it is possible to specifically amplify transposon junction fragments from SB-induced tumors using LM-PCR, pool the resulting products, and sequence them in a single run on the GS FLX System.
We obtained adequate genomic DNA samples from 17 skin (squamous cell carcinoma) and 11 liver (hepatocellular carcinoma) tumors from T2/Onc3 mice, and these samples were analyzed using LM-PCR followed by pyrosequencing. This approach generated 692 sequences on average for each tumor, with as few as 261 and as many as 1273 sequences per tumor. Overall, this represents a 3-fold increase over the sequence coverage previously used to identify insertion sites in SB-induced tumors (6). We developed an automated process to trim, map and annotate each sequence called the Integration Analysis System (http://ias.eng.uiowa.edu/IAS/).
We were able to identify an average of 166 unique integration sites in each SB-induced skin (Supplementary Table S4) and liver tumor (Supplementary Table S5). Most of the mapped integration sites (50–80%) were represented by a single sequence. It is possible that these sequences result from subclonal transposition events in rare tumor cells. Alternatively, these single sequences could arise from the contamination of normal cells that also undergo transposition in these mice due to ubiquitous transposase expression. In most tumors, however, a small number of integration sites contributed the majority of sequences. For example, an average of 45% of the sequences from each tumor were derived from only 10% of the integration sites (Supplementary Fig. S5). The clonal nature of these sites suggests they are the initiating transposon-induced mutations that gave rise to the tumor (Supplementary Tables S4 and S5).
Analysis of transposon integration site data in SB-induced skin and liver tumors
Over 1,800 and 3,000 integration sites were identified in SB-induced skin and liver tumors, respectively. We then defined common integration sites (CISs) by determining if specific genomic regions were disrupted by transposon integration at a significantly higher rate than expected. Previous SB mutagenesis screens have shown that such regions often contain cancer genes (5, 6). However, previous retroviral insertional mutagenesis screens have provided different definitions of a CIS (18, 19), raising the question of how best to determine their significance.
Significant CIS detection was recently improved by using Monte Carlo methods to simulate random integration in the host genome (11, 16). The simulated integration pattern can then be used to calculate the significance of any candidate CIS by determining the probability that integrations could occur by chance within a data set of a given size. Other methods developed to analyze integration patterns in lymphomas induced by retroviral insertional mutagenesis have recently been developed (20). However, a Monte Carlo method better simulates the specific features of SB transposition. For instance, the target sequence for SB transposon integration is a TA dinucleotide. The Monte Carlo simulation assumes that transposons will be randomly distributed at TA dinucleotide sites throughout the genome. One exception to this assumption is needed to account for local hopping since transposons tend to integrate in regions physically linked to the donor transgene. We determined the local chromosomes for TG12740 (chr9) and TG12775 (chr12) by analyzing the distribution of transposon integration sites in tumors (Supplementary Fig. S6). All integration events on the local chromosome were removed from the data set for each strain to eliminate false positive CISs due to local hopping. Significant CISs were then identified using criteria defined by Monte Carlo simulation (Supplementary Table S6). This method identified 3 CISs in skin and 2 CISs in liver tumors (Table 2).
Table 2
Table 2
Identification of CISs and CIGs using Monte Carlo simulation (α = 0.001).
The most common transposon-induced mutation in liver tumors was within the Rian locus, and integrations were distributed throughout the 5’ end of the gene with no apparent bias for transposon orientation (Fig. 5A). The ProTIS analysis tool indicates these integration sites are not predicated to be preferred TA sites (Supplementary Fig. S7) (21). These observations strongly suggest that transposon disruption of the Rian gene is selected during T2/Onc3-induced HCC. The Rian locus is complex and includes three known microRNAs and a C/D snoRNA cluster (22). Additional experiments are therefore required to determine how expression of these various elements is impacted by transposon insertion.
More than 80% of SB-induced skin tumors had a transposon insertion in the Zmiz1 gene. Zmiz1 encodes a 115 kD protein that contains a MIZ (MSX-interacting zinc finger) domain along with a putative nuclear localization signal and transcriptional activation domains (23). All 15 transposon integration events occurred in four different TA sites within intron 8 of Zmiz1 (Fig. 5B). Southern blot analysis confirmed the LM-PCR results in all cases, and identified additional SB-induced skin tumors with Zmiz1 integrations not analyzed by LM-PCR (Supplementary Fig. S8). All 15 integrations are in the same transcriptional orientation as Zmiz1 and are predicted to overexpress a fusion transcript between the CAG promoter in the T2/Onc3 transposon and exons 9–24 of Zmiz1. We confirmed the presence of a fusion transcript by RT-PCR (Fig. 5C). The fusion transcript is predicted to encode an N-terminally truncated protein. It is unclear how this truncation might affect Zmiz1 protein activity, as this N-terminal region does not contain any known functional motifs.
In this report we describe the generation of a Cre-inducible Sleeping Beauty transposon system that eliminates the embryonic lethality associated with the original SB system (6), and allows tissue-specific control of SB transposase expression, modifying the tumor spectra produced. In the absence of a Cre transgene, the SB transposase is not expressed and tumors are not generated. This allows us to maintain mouse strains that are double homozygous for the conditional RosaSBaseLsL allele and a mutagenic transposon donor transgene. These mice can be used to generate large cohorts of tumor-prone animals through a single cross to a tissue-specific Cre transgenic line, and have already been employed to generate models of colorectal cancer and hepatocellular carcinoma (11, 16). Here, we use an Aid-Cre allele to activate transposon mutagenesis in germinal center B-cells.
Finally, we show that altering the structure of the mutagenic SB transposon can have a profound effect on the tumor types induced by transposition. While both are ubiquitous promoters, MSCV is very efficient in hematopoietic cells and CAG works efficiently in epithelial cells (24). By replacing the MSCV LTR in the original T2/Onc2 transposon with the CAG promoter, we shifted the tumor spectrum from mostly lymphomas to primarily carcinomas. This suggests the promoter used to generate gain-of-function mutations in oncogenes upon SB integration significantly impacts the tumor types produced. This further implies that the tumor spectrum generated by SB mutagenesis could potentially be altered by placing tissue-specific promoters within the mutagenic transposon. When combined with the Cre-inducible RosaSBaseLsL allele, transposon promoter variants could provide even finer control of Sleeping Beauty mutagenesis in vivo.
The SB-induced carcinoma model described is highly significant, not only because of its ability to generate solid tumors, but also because of its ability to produce metastatic tumors. While the metastasis rate was low, it should be noted that each animal had an average of 3 independent tumors at the time of sacrifice, and high tumor burden may have caused the animals to become moribund before any macroscopic metastatic lesions could develop. Restricting SB mutagenesis to a specific tissue using the conditional RosaSBaseLsL allele could potentially increase the rate of metastasis. An alternative approach would be to induce SB mutagenesis in a transgenic or knockout strain background that develops spontaneous tumors with low metastatic potential, in order to identify mutations that accelerate metastasis. The SB system could then be used to study metastasis by identifying genes that contribute to a metastatic cell’s ability to survive and proliferate outside of the primary tumor site (25).
The power of forward genetic screens using SB transposition is greatly enhanced by larger tumor panels. Unfortunately, the types of carcinoma that developed in mice with the transposon carrying the CAG promoter were so diverse that we were unable to obtain significant numbers of tumors for most tumor types. However, we were able to identify novel candidate cancer genes for squamous cell carcinoma of the skin and hepatocellular carcinoma, the two most common tumor types in these mice. The limited number of samples is likely the reason we did not identify more candidate cancer genes in these two tumor types. The genomic landscapes of SB-induced tumors and human tumors are similar in that relatively few candidate cancer genes are mutated by SB transposon insertion in a large percentage of tumors. These genes can be easily identified in even small tumor panels (i.e. Zmiz1, Rian). Analysis of large SB-induced tumor panels will be required to identify candidate cancer genes that appear as gene hills on the cancer genomic landscape.
The SB system we describe here demonstrates the flexibility and power of this system for identifying novel cancer genes in many tumor types. Our work also suggests that SB transposition is able to model many forms of human cancer in mice, including carcinomas. Thus, the SB system offers a complementary approach in the identification of cancer genes and may play an important role in deciphering the genetic complexity of human tumors. The SB system will likely identify a set of candidate genes that partially overlaps those identified in human tumors. Genes identified by both approaches are much more likely to represent cancer genes and would merit further study. Additionally, the SB transposition system could be used to address basic biological questions, such as metastasis or resistance to therapeutic agents, that would be difficult to address directly in human cancer patients.
Generation of RosaSBaseLsL and T2/Onc3 mice
The RosaSBaseLsL targeting construct was generated by cloning three polyA signals (two SV40 polyA sites and one bovine growth hormone polyA) downstream of an EGFPF cDNA (Clontech). Flanking loxP and restriction sites were added to this amplified cassette to allow the fragment to be cloned between the En2 splice acceptor and SBase cDNA in the targeting construct used to produce the RosaSBase allele(5). A PGKneo cassette flanked by FRT sites was inserted downstream of the SBase-polyA sequences. The targeting vector was linearized and electroporated into hybrid C57BL/6J×129 ES cells. Following selection, clones were screened by Southern blotting. Chimeric offspring resulting from blastocyst injection were bred to C57BL/6J mice to establish the line.
The T2/Onc3 transposon was generated by replacing an EcoRI/HindIII fragment from the pT2/Onc2 plasmid(5) containing the MSCV 5’LTR promoter with a EcoRI/HindIII fragment containing the CAG promoter. The pT2/Onc3 plasmid was linearized by ScaI. Linear plasmid DNA was injected into C57BL/6J×C3H hybrid one-cell embryos. Tail-biopsy DNA from resulting offspring was screened by Southern blotting.
All animal experiments were approved and monitored by the NCI-Frederick Animal Care and Use Committee.
Immunohistochemistry and histopathology
The Comparative Molecular Pathology Unit at NCI-Frederick performed histopathology and tumor classification.
Statistical analysis of common integration sites
Monte Carlo and gene-based Monte Carlo methods are described in the supplementary methods.
Paraffin Block RNA Extraction and RT-PCR to detect Zmiz1 fusion transcripts
Total RNA was extracted from formalin-fixed, paraffin-embedded mouse tumor tissue using the RecoverAll Total Nucleic Acid Isolation Kit (Ambion). RT-PCR was performed with the OneStep RT-PCR Kit (Qiagen). Primer sequences were as follows: SD-F (5’-AAGCTTGCTACTAGCACCAGAACGCC-3’), SA-R (5’-GCTACGTTGCTAACA ACCAGTGCGG-3’), ex8-F (5’-TTTCCTTATGACTCTGTCCCTTGG-3’), ex9-R (5’-GTTG CTGGTGCTCATGCCTGATG-3’), mGAPDH-F (5’-GGTCCTCAGTGTAGCCCAAG-3’), and mGAPDH-R (5’-AATGTGTCCGTCGTGGATCT-3’).
Sequence analysis and annotation
GS FLX sequences were analyzed to identify the expected transposon/adaptor sequences using BLAST (26). The flanking genomic insert was extracted into a separate file. Sequences with the correct structure were sorted via barcode incorporated during PCR. The junction sequence was aligned to the mouse genome (mm9) using BLAT (27). Unique best hits were identified from the genome alignments. In cases of multiple alignments, the score of the second-best alignment was required to be at least 5% lower than the score of the best alignment. Integrations were aggregated for each tumor and annotated based upon the chromosome, basepair position and the number of sequence reads from that integration site. The UCSC genome annotation database refFlat table was used to determine if integration occurred within, or nearby, a gene (28). If so the gene symbol, relative location within the gene, expected effect on expression, and orientation with respect to the gene were annotated.
Monte Carlo simulation to identify significant common integration sites (CISs)
A modified Monte Carlo method generated a randomly selected TA site in the mouse genome for each transposon integration that had been identified. Next, 1x106 iterations of the simulation were performed. The output was used to define a CIS as the minimum genomic region in which N different insertions are observed to be significant (p=0.001) when compared to 1x106 iterations of the Monte Carlo simulation. We determined this value for CISs that contain between 3 and 10 insertions and for which a single tumor contributes no more than 2 of the events (Supplementary Table S5).
Figure 4
Figure 4
T2/Onc3 integration into the Rian and Zmiz1 loci
Supplementary Material
ACKNOWLEDGEMENTS
We thank D. Swing and R. Koogle for generating the T2/Onc3 transgenic mice and maintaining all mouse strains, and E. Southon and S. Reed for generating mice carrying the RosaSBaseLSL knock-in allele. The production and characterization of all mouse strains was supported by the Department of Health and Human Services, National Institutes of Health and the National Cancer Institute. The molecular characterization of tumors was supported by a grant from the American Cancer Society (IRG-77-004-28).
1. Wood LD, Parsons DW, Jones S, et al. The genomic landscapes of human breast and colorectal cancers. Science (New York, NY. 2007;318:1108–1113. [PubMed]
2. Jones S, Zhang X, Parsons DW, et al. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science (New York, NY. 2008;321:1801–1806. [PMC free article] [PubMed]
3. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme. Science (New York, NY. 2008;321:1807–1812. [PMC free article] [PubMed]
4. Ding L, Getz G, Wheeler DA, et al. Somatic mutations affect key pathways in lung adenocarcinoma. Nature. 2008;455:1069–1075. [PMC free article] [PubMed]
5. Collier LS, Carlson CM, Ravimohan S, Dupuy AJ, Largaespada DA. Cancer gene discovery in solid tumours using transposon-based somatic mutagenesis in the mouse. Nature. 2005;436:272–276. [PubMed]
6. Dupuy AJ, Akagi K, Largaespada DA, Copeland NG, Jenkins NA. Mammalian mutagenesis using a highly mobile somatic Sleeping Beauty transposon system. Nature. 2005;436:221–226. [PubMed]
7. Dupuy AJ, Jenkins NA, Copeland NG. Sleeping beauty: a novel cancer gene discovery tool. Hum Mol Genet. 2006;15(Spec No 1):R75–R79. [PubMed]
8. Collier LS, Largaespada DA. Hopping around the tumor genome: transposons for cancer gene discovery. Cancer Res. 2005;65:9607–9610. [PubMed]
9. Soriano P. Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat Genet. 1999;21:70–71. [PubMed]
10. Srinivas S, Watanabe T, Lin CS, et al. Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC developmental biology. 2001;1:4. [PMC free article] [PubMed]
11. Starr TK, Allaei R, Silverstein KA, et al. A Transposon-Based Genetic Screen in Mice Identifies Genes Altered in Colorectal Cancer. Science (New York, NY. 2009 [PMC free article] [PubMed]
12. Hawley RG, Fong AZ, Burns BF, Hawley TS. Transplantable myeloproliferative disease induced in mice by an interleukin 6 retrovirus. J Exp Med. 1992;176:1149–1163. [PMC free article] [PubMed]
13. Okabe M, Ikawa M, Kominami K, Nakanishi T, Nishimune Y. 'Green mice' as a source of ubiquitous green cells. FEBS Lett. 1997;407:313–319. [PubMed]
14. Niwa H, Yamamura K, Miyazaki J. Efficient selection for high-expression transfectants with a novel eukaryotic vector. Gene. 1991;108:193–199. [PubMed]
15. Bosch FX, Ribes J, Diaz M, Cleries R. Primary liver cancer: worldwide incidence and trends. Gastroenterology. 2004;127:S5–S16. [PubMed]
16. Keng VW, Villanueva A, Chiang DY, et al. A conditional transposon-based insertional mutagenesis screen for genes associated with mouse hepatocellular carcinoma. Nat Biotechnol. 2009 [PMC free article] [PubMed]
17. Gavin AJ, Scheetz TE, Roberts CA, et al. Pooled library tissue tags for EST-based gene discovery. Bioinformatics (Oxford, England) 2002;18:1162–1166. [PubMed]
18. Mikkers H, Allen J, Knipscheer P, et al. High-throughput retroviral tagging to identify components of specific signaling pathways in cancer. Nat Genet. 2002;32:153–159. [PubMed]
19. Suzuki T, Shen H, Akagi K, et al. New genes involved in cancer identified by retroviral tagging. Nat Genet. 2002;32:166–174. [PubMed]
20. de Ridder J, Kool J, Uren A, Bot J, Wessels L, Reinders M. Co-occurrence analysis of insertional mutagenesis data reveals cooperating oncogenes. Bioinformatics (Oxford, England) 2007;23:i133–i141. [PubMed]
21. Geurts AM, Hackett CS, Bell JB, et al. Structure-based prediction of insertion-site preferences of transposons into chromosomes. Nucleic acids research. 2006;34:2803–2811. [PMC free article] [PubMed]
22. Royo H, Cavaille J. Non-coding RNAs in imprinted gene clusters. Biology of the cell / under the auspices of the European Cell Biology Organization. 2008;100:149–166. [PubMed]
23. Sharma M, Li X, Wang Y, et al. hZimp10 is an androgen receptor co-activator and forms a complex with SUMO-1 at replication foci. The EMBO journal. 2003;22:6101–6114. [PubMed]
24. An DS, Wersto RP, Agricola BA, et al. Marking and gene expression by a lentivirus vector in transplanted human and nonhuman primate CD34(+) cells. Journal of virology. 2000;74:1286–1295. [PMC free article] [PubMed]
25. Nguyen DX, Massague J. Genetic determinants of cancer metastasis. Nature reviews. 2007;8:341–352. [PubMed]
26. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. Journal of molecular biology. 1990;215:403–410. [PubMed]
27. Kent WJ. BLAT--the BLAST-like alignment tool. Genome research. 2002;12:656–664. [PubMed]
28. Karolchik D, Baertsch R, Diekhans M, et al. The UCSC Genome Browser Database. Nucleic acids research. 2003;31:51–54. [PMC free article] [PubMed]
29. Donsante A, Miller DG, Li Y, et al. AAV vector integration sites in mouse hepatocellular carcinoma. Science (New York, NY. 2007;317:477. [PubMed]