PB somatic mutagenesis with an activated luciferase reporter
In order to develop a somatic mutagenesis system that can induce mutations while simultaneously and non-invasively (ex vivo
) reporting mutagenesis levels in tissues, we engineered a strategy to bioluminescently mark cells in which transposon mobilization has occurred. We took advantage of the highly sensitive luciferase reporter gene 
by inserting PB[mut]
at a TTAA tetranucleotide target site within the coding sequence (, Luc-PB[mut]
), thereby preventing luciferase expression from an upstream Actin promoter. During mutagenesis, active PB transposase (PBase) catalyzes the precise excision of the PB mutator transposon and a full-length luciferase product is reconstituted (). Five Luc-PB[mut]
transgenic lines were established and their ability to report PBase activity was tested by crossing to our previously described Act-PBase
. The Luc-PB[mut]7
line was identified as the most robust reporter for PB mutagenesis activity that can be easily assayed by ex vivo
imaging (). Luc-PB[mut]7
behaves as a faithful reporter for PBase activity, since it does not produce any luciferase signal in mice lacking PBase ().
To examine whether Luc-PB[mut]7
can differentiate between varying levels of PBase activity, we generated transgenic mice expressing a PBase-Estrogen Receptor fusion transgene (Act-PBaseER
). In the inactive state, the PBaseER protein is sequestered in the cytoplasm. Stimulation with tamoxifen shuttles PBaseER to the nucleus and activates its ability to drive PB transposition (Figure S1
). In vivo
, luciferase signal was not observed in newly weaned, untreated Luc-PB[mut]7;Act-PBaseER
transgenic mice. However, transgenic Act-PBaseER
can mobilize PB upon tamoxifen administration () and luciferase signal can be detected one week after treatment, providing a tool for conditional PB TIM. Quantitative PCR measuring PB[mut]
excision revealed that PBaseER activity levels were significantly lower than those of Act-PBase
(). These results were corroborated by the luciferase mutagenesis reporter signal from ex vivo
imaging (). Furthermore, we sometimes observed leaky luciferase signal in older, untreated mice housed together with tamoxifen-treated mice (data not shown). This signal, however, is at least an order of magnitude lower than signal from treated mice as measured by ex vivo
imaging (data not shown). The ability to differentiate between differing levels of luciferase signal indicates that Luc-PB[mut]7
is a sensitive reporter for PBase activity.
By real-time PCR analysis, the Luc-PB[mut]7 transgenic line was found to contain seven copies of the mutagenic transposon. When crossed to Act-PBase mice, the number of live-born double transgenic pups was 93% of the expected number (n>100). Since mutagenesis in Luc-PB[mut]7;Act-PBase mice was observed in all tissues as reported by luciferase, we addressed whether this low copy line is sufficient to induce different phenotypes when mutagenesis is targeted in all tissues or specific organs. We found Luc-PB[mut]7 can induce tumor formation as well as other phenotypes including cellular features (e.g., cell infiltration and clonal expansion) and morphological defects (e.g., alopecia) (also see below).
We found that body-wide PB-SMART can induce tumor formation in non-sensitized backgrounds. We observed that Luc-PB[mut]7;Act-PBase
mice developed tumors while conducting sensitized screens. Nine out of 32 of these Luc-PB[mut]7;Act-PBase
) mice developed tumors within 50 weeks while only one out of seventeen Act-PBase
) and zero out of 32 Luc-PB[mut]
) controls developed a tumor (). We have collected additional tumors from cohorts that have not reached 50 weeks and other non-sensitized cohorts containing floxed latent alleles of Pten
loss or Braf
. Across these cohorts, we have observed similar tumor types including kidney and pancreatic tumors, lung adenocarcinoma, hepatocellular carcinomas, soft tissue sarcomas such as angiosarcoma and small round blue cell (SRBC) tumors, and skin tumors such as squamous cell carcinoma, sebaceous cancer, and melanoma. These results confirm that our low copy PB-SMART system can efficiently induce tumor-promoting mutations in a wide variety of tissues.
We also found that the low copy mutagenic transposon in the PB-SMART system allows for easy identification of potential causative insertional mutations. Transposon insertions from tumors were mapped by linker-mediated PCR (LM-PCR). On average, across 45 tumors analyzed, 6.1 PB insertions were mapped. Notably, greater than 64% of all insertions were recovered in introns, exons, or within ten kilobase pairs (kb) of either known or predicted genes (). Given the propensity of Luc-PB[mut] for inserting into or near genes in these tumors, we analyzed whether the mutator transposon was oriented along or opposite the coding direction of inserted genes. Because our mutator transposon is designed to overexpress full-length genes or truncated genes, we expected that in tumors, insertions oriented along the coding direction of nearby or affected genes would be enriched versus insertional orientation by random chance. Indeed, over 64% of intron insertions were oriented along the coding direction of the affected genes, and almost 60% of insertions within 10 kb upstream of genes were oriented toward the coding direction of the nearby genes (). Meanwhile, insertions within 10 kb downstream of nearby genes displayed less bias for mutator orientation (). These data imply that the PB mutator transposon is indeed upregulating genes important for tumorigenesis.
Insertional frequency in tumors induced by Luc-PB[mut]7 and PB[mut-RFP]7.
Mutator orientation in tumors induced by Luc-PB[mut]7 and PB[mut-RFP]7.
In five kidney tumors, we identified common insertions in Mitf. The mutator transposons in all five tumors were inserted in the coding direction, upstream of the translational start of the M-isoform, suggesting overexpression of Mitf in these tumors (). Indeed, quantitative PCR revealed that Mitf was upregulated in these kidney tumors as compared to two wild-type kidney samples (). Histopathology of the kidney tumors revealed an expansion of nests of pleomorphic spindle and epithelioid cells suggestive of carcinoma ().
Identification of driver oncogenes in solid tumors from Luc-PB[mut]7;Act-PBase mice.
Notably, two other members of the MiT family of basic-helix-loop-helix/leucine zipper transcription factors, TFE3
, have been recurrently affected by translocations that result in the upregulation of these transcription factors in human pediatric renal cell carcinoma 
. Since TFE3, TFEB, and MITF are highly homologous and bind to the same DNA consensus sequence, it is likely that they can substitute for each other as oncogenes 
. Together, these data suggest that MITF
upregulation is an important driver for kidney tumor formation.
We also identified Gli2
, a downstream effector of the Hedgehog
) signaling pathway 
, as a common insertion site in SRBC tumors. We mapped coding-direction insertions in intron 7 or intron 8 from 11 tumors displaying SRBC morphology (). Gli2 contains a repressor domain in the amino-terminus, and expression of the C-terminal portion of Gli2 has previously been shown to result in the constitutive transcriptional activation of downstream target genes 
. In these sarcomas, it is likely that the Gli2
insertions lead to the expression of constitutive transcriptional activator forms of Gli2 that drive tumor formation. In fact, constitutive Hedgehog signaling in Gorlin's syndrome due to mutation in PTCH
leads to an increased propensity to develop an SRBC tumor, rhabdomyosarcoma 
. Furthermore, activation of Gli1 (a Gli2 transcriptional target) seems to be a major downstream effector of the EWS-FLI1 oncoprotein which drives another SRBC tumor, Ewing Sarcoma 
. Together these data strongly support the notion that constitutive Hedgehog signaling is a critical driver in SRBC tumor formation, with overexpression of truncated Gli2
being one mechanism. Thus, the identification of cancer genes with known human relevance validates the utility of our low-copy mutator for uncovering disease genes.
PB somatic mutagenesis with an activated RFP cell tracker
In somatic mutagenesis, the ability to track mutated cells via the same transgenic line used to induce mutations opens up many experimental possibilities. We thus added new elements in our PB mutator transposon to label the mutant cells. The RFP coding sequence and an internal ribosome entry site (IRES) were inserted between the Actin promoter and myc-SD, which contains start codons in all three reading frames followed by a splice donor, of PB[mut]
). Following insertions into introns or 5′ regulatory regions, production of bicistronic pre-mRNA containing RFP, IRES, the engineered myc-SD exon, and downstream endogenous intron and exon sequences can be initiated by the Actin promoter. After splicing of the myc-SD exon to downstream endogenous splice acceptors, the bicistronic transcripts enables the expression of RFP and the ectopic expression of endogenous proteins (). When we tested this mutator in cultured cells, the activated RFP marker was co-expressed in cells that ectopically expressed downstream endogenous proteins that incorporated the engineered myc-SD exon (). In order to ensure that PB[mut-RFP]
does not express RFP from a transgenic concatamer, a concatamer silencer (CS) was inserted outside of the transposon (). The CS contains a splice acceptor followed by stop codons in all three reading frames to terminate transcripts initiated by the Actin promoter from transgene concatamers. To prevent alternative splicing–mediated exclusion of the CS splice acceptor, a sequence forming a pre-tRNA-like (ΔAC) structure was cloned after the stop codons. This substrate is efficiently recognized and cleaved by RNase P ensuring that pre-mRNA which contain the CS do not form mature transcripts 
Simultaneously inducing mutations and tracking mutated cells with a PB transposon system.
We generated seven transgenic lines varying in transposon copy number from two copies to 200 copies () and found that none were RFP-positive, indicating that the CS successfully prevents undesired RFP production. When the two highest copy lines were crossed to the Act-PBase line, live-born double transgenic progeny were rarely observed (), suggesting that the mutagenic transposons can be mobilized by Act-PBase during embryonic development and can induce significant deleterious effects and embryonic lethality. The mobilization of the low copy PB[mut-RFP] lines was also confirmed by excision PCR (). Importantly, mutagenesis using either of the two-copy or the seven-copy PB[mut-RFP] lines crossed to Act-PBase mice resulted in the appearance of RFP-positive patches () and triggered tumorigenesis (), validating these three low copy PB[mut-RFP] lines for somatic mutagenesis with an activated cell tracker.
Tissue-specific PB somatic mutagenesis screening with an activated reporter and tracker (PB-SMART screen)
We have shown that the collective luciferase signal in tissues generated by Luc-PB[mut]7 can act as a robust reporter of mutagenesis activity. Since Luc-PB[mut]7 is also a mutator, the luciferase signal could also be used to mark the mutant cells. Given that Luc-PB[mut]7 is a low copy line and not every cell in a tissue will be luciferase positive, we reasoned that clones or patches of mutant cells will have elevated signals and could be detected through ex vivo imaging. In this sense, the luciferase reporter also behaves as a tracker for mutant cells. By monitoring ex vivo luciferase signal, we identified clonal patches in the skin and tracked the appearance of melanomas in Luc-PB[mut]7;Act-PBase;BrafCA;Tyr-CreER mice ( and ; also see below). In addition, the luciferase signal enabled the identification of tumor cell infiltration into distant organs of a Luc-PB[mut]7;Act-PBase;Pten+/- mouse that developed lymphoma (). The enlarged thymus and lymph nodes were strongly luciferase-positive due to the dense clustering of cells that activated the luciferase reporter (). Strong signal was also seen emanating from a distinct portion of the liver (). Histology revealed that the luminescent spot was indeed infiltration of lymphocytes (), and we mapped identical transposon insertions from the liver infiltration and other tumor sites, indicating that the cells originated from the same clone. Thus, the Luc-PB[mut]7 line is a mutant clone tracker for monitoring clonal expansion, tumor formation, and infiltrations of cells into foreign tissues.
Tracking tumor formation and infiltration with Luc-PB[mut]7.
Tissue-specific induction of skin phenotypes by Cre-activated PB mutagenesis.
We further expanded the PB-SMART system for tissue-specific genetic screens. To allow the system to be readily adopted for screens in many types of tissues and cells, we generated transgenic mice in which PBase
can be activated by tissue-specific Cre by inserting a Stop-pA sequence flanked by loxP sites between the Actin promoter and the transposase coding sequence (, LSL-PBase, LSL-PBaseER
). When these double transgenics were crossed into lines expressing Cre in the kidney and genitourinary tract or the skin epithelium 
, correct targeting of mutagenesis to these tissues was confirmed by the luciferase reporter ().
To ask whether Luc-PB[mut]7
can be used to track mutation-driven expansion of cells, we utilized the Tyr-CreER
line, which expresses CreER specifically in melanocytes 
, one of the least densely clustered cell types in the body. Tyr-CreER
has been used to conditionally activate BrafV600E
, thereby promoting melanocyte proliferation 
. We generated Luc-PB[mut]7;LSL-PBase;BrafCA;Tyr-CreER
quadruple transgenic mice and assayed whether luciferase can track mutation-driven expansion of cells ex vivo
. After administering tamoxifen, we were able to detect luciferase signal in melanocytes in the ears, tail, and dorsal trunk and also track the expansion of the positive cells over time ().
To further illustrate that Luc-PB[mut]7
can be used to track clonal expansion in specific cell types, we targeted the PB-SMART system to the mouse skin. We generated Luc-PB[mut]7;LSL-PBase;K14-Cre
mice which express Cre specifically in the basal keratinocytes of the epidermis. This compartment contains the stem and progenitor cells that give rise to the hair follicle, sebaceous gland, and interfollicular epidermis 
. We reasoned that if a keratinocyte stem or progenitor cell were to be targeted by PB-SMART, over time an entire clonal unit would become luciferase positive. Indeed, the appearance of distinct luciferase clusters could be detected in these mice ().
Skin-specific PB-SMART resulted in tumor development as well as other morphological phenotypes. For example, we observed hair loss on dorsal skin, resulting from PB mutagenesis and marked by luciferase signal (). Keratin staining of frozen sections revealed a striking epidermal thickening in the affected region (). We mapped a transposon insertion upstream of Myc
in the affected skin (). This is consistent with previous reports that in K14-c-Myc
transgenic mice, epidermal stem cell proliferation and differentiation results in hair loss, hyperproliferation, and thickening of the interfollicular epidermis 
We further analyzed skin tumors that were induced and labeled by skin-specific PB-SMART. We were able to observe tumors in Luc-PB[mut]7;LSL-PBase;K14-Cre
mice starting at four months of age. Using the ex vivo
reporter system, we can detect regions of luciferase activity prior to tumor formation, that correspond to the later development of tumors (). In four of these tumors, we identified transposon insertions in Gli2
(). Quantitative PCR confirmed that Gli2
mRNA levels were increased in these tumors as compared to wild type skin (). Constitutive activation of the Hh
pathway is characteristically seen in basal cell carcinoma (BCC) 
, which arises from stem cells in the epidermis and hair follicle 
. Furthermore, histopathology of the skin tumors revealed palisading basaloid cells forming tumor nests and retraction of the stroma from the tumor islands in formalin-fixed sections similar to human BCC (). Together, these data show that the PB-SMART system can be deployed to conduct tissue-specific genetic screens and identify causative mutations relevant to human disease.