Sleeping Beauty transposon insertional mutagenesis based mouse models for cancer gene discovery

Abstract

Large-scale genomic efforts to study human cancer, such as the cancer gene atlas (TCGA), have identified numerous cancer drivers in a wide variety of tumor types. However, there are limitations to this approach, the mutations and expression or copy number changes that are identified are not always clearly functionally relevant, and only annotated genes and genetic elements are thoroughly queried. The use of complimentary, nonbiased, functional approaches to identify drivers of cancer development and progression is ideal to maximize the rate at which cancer discoveries are achieved. One such approach that has been successful is the use of the Sleeping Beauty (SB) transposon-based mutagenesis system in mice. This system uses a conditionally expressed transposase and mutagenic transposon allele to target mutagenesis to somatic cells of a given tissue in mice to cause random mutations leading to tumor development. Analysis of tumors for transposon common insertion sites (CIS) identifies candidate cancer genes specific to that tumor type. While similar screens have been performed in mice with the PiggyBac (PB) transposon and viral approaches, we limit extensive discussion to SB. Here we discuss the basic structure of these screens, screens that have been performed, methods used to identify CIS.

Development of Sleeping Beauty (SB) and general characteristics

SB is a two-component system comprised of a DNA element termed the transposon vector and an enzyme component, termed the transposase, originally constructed in Ivics et al. [1••]. When these two components are present within the same cell, the transposase recognizes the inverted repeat/direct repeat (IR/DR) sequences flanking the transposon and catalyzes the “cut-and-paste” reaction. In this process, the mobilized transposon vector is excised from the donor locus and reintegrated elsewhere at a TA dinucleotide. Several improvements to the SB system were made early on, including changes to the inverted terminal repeats that increase transposition rates [2]. Secondly, various catalytically improved versions of the SB transposase have been produced by site directed mutagenesis. Because the first active version of the SB transposase was called SB10 [1••], improved versions were called SB11 [3] and the like. By far the most active version is called SB100, and was the result of extensive screening of randomly generated derivatives of earlier versions [4].

Early studies using SB for germline mutagenesis and transgenesis identified important features of SB transposition that have relevance for somatic cell SB transposon mutagenesis. It was found transposition rates from multicopy transposon concatemers far exceeds the rate at which single-copy transposon vectors can be mobilized, even accounting simply for the increase in the number of substrates for transposition [5–7]. The reasons for this are not entirely clear, but it is known methylated SB transposon vector DNA is a better substrate for transposition compared to unmethylated transposon [8], and many transgene concatemers are partially methylated. A second important feature is called “local hopping.” Local hopping refers to the tendency of transposons to land near the donor concatemer on the same chromosome, usually within ~2–10 megabase pairs (Mb). This tendency is clear with SB transposition from donor concatemers located within mouse chromosomes for both germline and somatic cell transposition [5,7,9••,10••].

SB mediated insertional mutagenesis for cancer gene discovery

The SB system was first used for mutagenesis via a body wide screen for cancer in mice expressing the SB transposase (SB10 or SB11) and a mutagenic transposon line (T2/Onc or T2/Onc2) [9••,10••]. Tumors resulted from insertion mutations in or near endogenous genes. In this work, loci recurrently mutated by transposons more than expected by chance, (that is, in multiple independent tumors) called common insertion sites (CIS), were identified. The T2/Onc transposons were designed to induce either gain-of-function (GOF) or loss-of-function (LOF) mutations when inserted in or near a gene based on its genetic cargo. The murine stem cell virus (MSCV) long terminal repeat (LTR) promoter with artificial exon and splice donor (SD) was included so downstream exons could be ectopically overexpressed as a consequence of fusion with transcripts initiated by the LTR and splicing from the T2/Onc SD. Many examples of ectopic overexpression of proto-oncogenes via this mechanism have now been described. In some cases, the fusion transcript produced encodes a full length protein as the targeted gene encodes a protein which has a translational start site in exon 2, for example in the case of activation of Rspo2 [11••,12]. In other cases, insertion within a gene is followed by production of LTR initiated transcripts, splicing from the SD, to make fusion transcripts that encodes an N-terminally truncated protein translated from an internal ATG start codon, such as seen with Notch1 [10••] and Braf [9••]. The T2/Onc vectors also included splice acceptors in both orientations and a bidirectional polyadenylation signal, to terminate transcripts that splice into the vector after insertion within an intron of a gene. In this way, many tumor suppressor genes (TSGs) have been inactivated as a consequence of SB insertional mutagenesis in various screens. Transcript termination has also resulted in proto-oncogene activation by C-terminal truncation, as in the case of Egfr [13••]. Production of competitive endogenous RNAs (ceRNA), resulting in TSG downregulation via microRNAs, has also been observed as a consequence of SB transposon insertion [14].

One important variation to T2/Onc structure was made in version 3, termed T2/Onc3, in which the MSCV LTR was replaced by the CMV enhancer/chicken beta-actin (CAG) promoter [15•]. CAG has reduced activity in hematopoietic cell types and enhanced activity in epithelial cell types [15•]. Evidence suggests that T2/Onc3 may more potently induce tumors in epithelial tissues compared to T2/Onc and T2/Onc2, and lead to activation of proto-oncogenes more readily. The development of T2/Onc3 demonstrates that changes to the structure of SB transposons used for mutagenesis could reveal new kinds of genes and genetic elements in cancer development than have been discovered in screens so far.

It is important to consider the likelihood that SB-induced or accelerated tumors in mice have mutations caused by mechanisms other than T2/Onc insertional mutagenesis. Indeed, copy number alterations have been observed in tumor cells from SB cancer models [16•,17]. Nevertheless, SB-induced sarcomas generally have greatly reduced whole chromosome and gene copy number changes compared to sarcomas induced in mice without SB mutagenesis [16•]. The extent to which noninsertional mutagenesis mechanisms, such as point mutations, translocations, deletions and amplifications, contribute to SB models of cancer in mice remains to be determined.

The first SB mutagenesis studies generated both blood and solid cancers and identified CIS-associated genes [9••,10••]. The logical progression of the approach led to the development of a conditional SB transposase allele (Rosa26-LSL-SB11) that could be activated in a desired tissue when combined with tissue specific Cre recombinase transgene [11••,13••,15•]. This allows for tissue specific SB mutagenesis. The overall structure of most SB screens is shown in Figure 1. Since its inception the conditional SB transposon mutagenesis system has been applied to many cancer types, identifying hundreds of candidate cancer genes, generating new cancer models, and providing insights into the genes and mechanisms of cancer progression.

Figure 1.

Basic structure of a SB cancer screen. Most screens to date utilize mice carrying the conditional SB11 Rosa26 “knockin” allele (Rosa26-LSL-SB11) and T2/Onc concatemer, which are crossed to mice carrying a tissue specific Cre transgene (TSP-Cre) and in many cases some cancer predisposing mutation (e.g. LSL-Trp53^R270H). Important choices to be made at each step of such a screen are boxed.

SB screens carried out so far

To date there have been dozens of SB screens performed for various cancers on many different predisposing backgrounds, which have identified numerous known and novel cancer genes (Table 1). These data demonstrate many different tissues are amenable to SB mutagenesis. These include tissues derived from all three germ layers, resulting in carcinomas, sarcomas, neuroectodermal tumors, as well as hematopoietic cancers. Many of the most common and most deadly forms of human cancer have been modeled using SB mutagenesis.

Table 1.

Sleeping Beauty (SB) based screens for cancer genes in vivo published to date. Beginning with the oldest papers, the first author, year of publication and title are shown as well as primary tissue/cancer type targeted in the work. References [9••,10••,11••,13••,14,16•,17,18,19•,20,21,23,27–47]

Author/year	Year	Cancer Type	Title
Dupuy et al.,	2005	Blood	Mammalian mutagenesis using a highly mobile somatic Sleeping Beauty transposon system.
Collier et al.,	2009	Blood	Whole-body sleeping beauty mutagenesis can cause penetrant leukemia/lymphoma and rare high-grade glioma without associated embryonic lethality.
Keng et al.,	2009	Liver	A conditional transposon-based insertional mutagenesis screen for genes associated with mouse hepatocellular carcinoma.
Starr et al.,	2009	Colorectal	A transposon-based genetic screen in mice identifies genes altered in colorectal cancer.
Berquam-Vrieze et al.,	2011	Blood	Cell of origin strongly influences genetic selection in a mouse model of T-ALL.
Collier et al.,	2011	Sarcoma	Cancer gene discovery in solid tumors using transposon-based somatic mutagenesis in the mouse.
Karreth et al.,	2011	Skin	In vivo identification of tumor-suppressive PTEN ceRNAs in an oncogenic BRAF- induced mouse model of melanoma.
Koudijs et al.,	2011	Blood	High-throughput semiquantitative analysis of insertional mutations in heterogeneous tumors.
March et al.,	2011	Colorectal	Insertional mutagenesis identifies multiple networks of cooperating genes driving intestinal tumorigenesis.
Starr et al.,	2011	Colorectal	A Sleeping Beauty transposon-mediated screen identifies murine susceptibility genes for adenomatous polyposis coli (Apc)-dependent intestinal tumorigenesis.
van der Weyden et al.,	2011	Blood	Modeling the evolution of ETV6-RUNX1-induced B-cell precursor acute lymphoblastic leukemia in mice.
Keng et al.,	2012	Liver	Sex bias occurrence of hepatocellular carcinoma in Poly7 molecular subclass is associated with EGFR.
Koso et al.,	2012	Nervous system	Transposon mutagenesis identifies genes that transform neural stem cells into glioma-initiating cells.
Mann et al.,	2012	Pancreas	Sleeping Beauty mutagenesis reveals cooperating mutations and pathways in pancreatic adenocarcinoma.
O’Donnell et al.,	2012	Liver	A Sleeping Beauty mutagenesis screen reveals a tumor suppressor role for Ncoa2/ Src-2 in liver cancer.
Pérez-Mancera et al.,	2012	Pancreas	The deubiquitinase USP9X suppresses pancreatic ductal adenocarcinoma.
van der Weyden et al.,	2012	Blood	Increased tumorigenesis associated with loss of the tumor suppressor gene Cadm1.
Wu et al.,	2012	Nervous system	Clonal selection drives genetic divergence of metastatic medulloblastoma.
Genovesi et al.,	2013	Nervous system	Sleeping Beauty mutagenesis in a mouse medulloblastoma model defines networks that discriminate between human molecular subgroups.
Lastowska et al.,	2013	Nervous system	Identification of a neuronal transcription factor network involved in medulloblastoma development.
Quintana et al.,	2013	Skin	A transposon-based analysis of gene mutations related to skin cancer development.
Rahrmann et al.,	2013	Nervous system	Forward genetic screen for malignant peripheral nerve sheath tumor formation identifies new genes and pathways driving tumorigenesis.
Tang et al.,	2013	Blood	Transposon mutagenesis reveals cooperation of ETS family transcription factors with signaling pathways in erythro-megakaryocytic leukemia.
van der Weyden et al.,	2013	Blood	Jdp2 downregulates Trp53 transcription to promote leukaemogenesis in the context of Trp53 heterozygosity.
Zanesi et al.,	2013	Blood	A Sleeping Beauty screen reveals NF-kB activation in CLL mouse model.
Rogers et al.,	2013	Mixed	Adaptive immunity does not strongly suppress spontaneous tumors in a Sleeping Beauty model of cancer.
Bard-Chapeau et al.,	2014	Liver	Transposon mutagenesis identifies genes driving hepatocellular carcinoma in a chronic hepatitis B mouse model.
Been et al.,	2014	Sarcoma	Genetic signature of histiocytic sarcoma revealed by a sleeping beauty transposon genetic screen in mice.
Vyazunova et al.,	2014	Nervous system	Sleeping Beauty Mouse Models Identify Candidate Genes Involved in Gliomagenesis
Koso et al.,	2014	Nervous system	Identification of FoxR2 as an oncogene in medulloblastoma.
Takeda et al.,	2015	Colorectal	Transposon mutagenesis identifies genes and evolutionary forces driving gastrointestinal tract tumor progression.
Perna et al.,	2015	Skin	BRAF inhibitor resistance mediated by the AKT pathway in an oncogenic BRAF mouse melanoma model.
Mann et al.,	2015	Skin	Transposon mutagenesis identifies genetic drivers of BrafV600E melanoma.
Moriarity et al.,	2015	Bone	A Sleeping Beauty Forward Genetic Screen Identifies New Genes and Pathways driving Osteosarcoma Development and Metastasis.

Sleeping Beauty mouse Models Identify Candidate Genes Involved in Gliomagenesis, PLoS One 2014 (Pubmed ID: 25423036).

We have learned from our own experiences, however, that not all cell types subjected to SB mutagenesis are equally amenable to tumor induction. In some of our work, the yield of tumors is low and latency long (data not shown). Thus we, and others, often use cancer prone genetic backgrounds in order to increase the tumor penetrance and decrease tumor latency. Tissue specific expression of oncogenes, such as Kras^G12D [18] and high level EGFR expression [17], and tissue specific or whole body TSG mutations, e.g. Ptch [19•] and Trp53 mutations [17], have been widely implemented. In many or perhaps most cases, SB induced acceleration of cancer on a predisposed background is the only method to obtain a high yield of tumors that develop within a reasonable length of time. Thus, care must be taken to choose the right version of the mutagenic SB transposon, Cre transgene, and cancer predisposed background (Figure 1). In any case, many reported models demonstrate the full spectrum of cancer development, including early benign lesions and malignant derivatives, including invasive and metastatic disease [16•,20].

Interestingly, there is a large difference in the frequency at which candidate oncogenes are identified compared to TSG between screens that modeled solid or hematopoietic cancer types. The frequency of oncogene candidates identified in hematopoietic cancers is much higher than observed in solid tumor screens, in which TSG candidates dominate. This is probably because most screens performed to date have used T2/Onc or T2/Onc2, which harbor the MSCV LTR, capable of efficient proto-oncogene activation in hematopoietic cells, but less so in other cell types. Thus, use of T2/Onc or T2/Onc2 lines may favor TSG inactivation as a mechanism of tumor induction during solid tumor development. Use of T2/Onc3 in more tissues may reveal additional proto-oncogenes in new screens. However, it is clear T2/Onc3 also disrupts TSG in screens reported so far [15•].

Current and future refinements of the approach

Several improvements, new methods of analysis, and applications of in vivo, somatic cell, transposon-based, insertional mutagenesis can now be envisaged or are underway (Figure 2). Indeed, the long and storied history of forward genetic screening in model organisms is marked by innovation in approaches and more and more complex screens. While initial SB work simply screened for cancer drivers in wild type mice, this was soon followed by screens for cancer genes and genetic pathways that cooperate with specific predisposing mutations [21,22]. Screens for specific cancer relevant traits, such as therapy resistance or metastasis could be very revealing. Indeed, several SB cancer models demonstrate overt metastasis [13••,16•] or invasive behavior [20] in a subset of cases and these studies have already revealed candidate genes whose alterations may cause these phenotypes. A recent study also showed that SB could be used to discover mediators of Braf inhibitor resistance in a model of melanoma in mice [23]. In a broader sense, SB models of cancer provide an opportunity to more realistically approximate a set of cancer cases like those that confront physicians in human patients. SB models provide a set of tumors with overlapping but highly complex mutations, which are tractable by analyses of insertion mutations. They also have intratumoral genetic heterogeneity, due to ongoing SB transposon mobilization as the tumor develops, a phenomenon that has been demonstrated in many types of human cancer [24]. These features could be used to help define genotype-phenotype correlations using SB models. The results would be especially relevant if these mouse models could be treated using therapies that mimic those given to human patients. In such a scenario, perhaps cancer cell autonomous determinants of treatment outcome could be discovered.

Figure 2.

Potential future refinements to SB cancer screens. Improvements and refinements to in vivo cancer screening using SB might be made at the level of the mouse, analysis of the primary tumor, or single cells isolated from the primary tumor.

Other methodological approaches to analyze SB mutagenized tumors should also be considered. For example, most SB screens use ligation-mediated polymerase chain reaction (LM-PCR) based recovery of transposon insertion sites from tumor DNA [25]. However, almost all screens to date use restriction enzymes as a first step in the LM-PCR protocol. This leads to a bias in the recovery of specific insertions due to uneven distribution of such sites. In fact, it has been demonstrated that shearing genomic DNA, before LM-PCR, allows for better recovery of insertions and leads to a good correlation between the number of times a given insertion is sequenced and the clonal abundance of cells with that insertion [26]. Thus, using this approach for insertion site recovery one could determine the “trunk” mutations that probably occurred very early in tumor development, versus the “branch” mutations that probably occurred later in tumor progression. Potentially, a different set of genes is characteristic for each kind of mutation. Another useful technology for such analyses would be the ability to recover transposon insertions from single cells. Generation of cell lines from SB induced tumors may allow specific phenotypes to be correlated with genotype more easily also.

Many SB screens are performed with two different T2/Onc (or T2/Onc2 or T2/Onc3) transgenic lines. In this way, it is possible to disregard local hopping and consider only insertion sites not linked to the donor concatemer but still cover the entire genome. To avoid using multiple concatemer lines, methods to eliminate local hopping in SB screens could be tried, such as addition of an artificial chromosome bearing the concatemer, or introduction of SB transposon vectors from plasmids, viral vectors, or episomal vector. Such approaches may have some practical value.

It is highly probable that complementary genomic analyses of SB induced tumors will be valuable. For example, it seems probably that some spontaneous genetic alterations cooperate with transposon insertion mutations in the development of these tumors. Therefore, array comparative genome hybridization, exome or whole genome sequencing and RNA sequencing may be revealing. RNA sequencing could be especially useful, because not only would the tumor gene expression profile (GEP) be revealed, but fusion transcripts between T2/Onc sequences (SA and SD) and endogenous transcripts would also be revealed. This would allow a focus on the most clonally abundant insertions, more easily determine whether a specific insertion was generating an LOF or GOF effect, and determine the target gene affected in the case of insertions within tightly clustered genes or for insertions some distance from their target gene.

Regardless of the details used to finally define all the CIS-associated genes, once a list of candidates is generated it can be used as an enriched list of tumor specific cancer genes for bioinformatic and comparative genomics analyses. For example, data from The Cancer Genome Atlas (TCGA), and similar projects, are becoming very valuable for comparing the specific genes and genetic pathways altered in both the SB mouse model and the corresponding form of human cancer [11••,27]. Such overlap provides strong evidence that a given gene or genetic pathway alteration is truly a cancer driver. However, functional validation in additional mouse models or human cells remains vital for making these determinations.

Concluding remarks

The development of SB and its use for cancer gene discovery has been a boon for cancer functional genomics studies. It has led to the discovery of new and specific cancer drivers critical to human tumor development or maintenance such as Rspo2, Foxr2, Sema4d, and others. The approach has also uncovered roles for specific cancer pathways in human cancer such as Pten and Wnt/beta-catenin regulated pathways in malignant peripheral nerve sheath tumors. However, the full impact of the technology on our understanding of cancer has yet to be realized. This is in part because the specific genes and pathways discovered are so numerous that the rest of the field has not yet focused attention on them. A second reason, is that the general approach has not been used to study specific cancer relevant phenotypes or for enhancer/suppressor screens that would give specific molecular insight into a given pathway or biological process. If adapted to more sophisticated screens of this kind SB has the potential to allow insight into the behaviors of rare, yet critical cancer cells, such as metastatic or treatment resistant cells. Moreover, it could be used to dissect in greater detail, pathways under selective pressure for change, in the unique in vivo environment of a normal tissue, a growing tumor, and in a metastatic niche.