A Versatile ES Cell–Based Melanoma Mouse Modeling Platform

Malignant melanoma is a cancer of melanocytes that is frequently fatal. Despite recent clinical advances in targeted and immune therapy, innate and acquired resistance to treatment necessitates the development of new therapeutic strategies. To this end, novel therapeutic targets must be identified to exploit melanoma-specific genetic dependencies and vulnerabilities. Genomic, genetic, and transcriptomic analyses have revealed numerous genes with putative functions in melanoma initiation, progression, and drug resistance (1–5). Extensive characterization of such candidate driver genes is required to assess their oncogenic potential, unravel their underlying molecular mechanisms, and examine opportunities for therapeutic targeting. Genetically engineered mouse models (GEMM) that faithfully recapitulate many aspects of the human malignancy are critical for the success of such efforts (6, 7).

Numerous melanoma GEMMs have been developed (8), many of which model genetic changes that are frequently observed in human melanoma. For instance, expression of oncogenic mutants of BRAF and NRAS (9–12), the most frequently mutated proto-oncogenes in melanoma (5), can be restricted to melanocytes when combined with a tamoxifen-inducible Cre allele driven by the tyrosinase promoter (13). Similarly, conditional knockout alleles of the predominant tumor suppressors in melanoma, PTEN, and CDKN2A (14, 15), can be specifically deleted in melanocytes by this approach. Combinations of these alleles result in mouse strains that develop melanoma with varying penetrance and latency (9, 10, 16, 17). As these GEMMs incorporate the most frequent alterations in human melanoma, they represent excellent “base” models to investigate the functions of additional candidate melanoma genes. Indeed, they have been used to examine the roles of mTorc1, Dnmt3b, Akt, and Trp53 in melanoma progression (18–21).

The functions of many genes with putative roles in melanoma remain unexplored. For instance, functional in vivo screens identified hundreds of candidate genes that may have oncogenic or tumor suppressive effects in melanoma (9, 22, 23). Similarly, CRISPR screens identified candidate genes having putative roles in melanoma drug resistance (24, 25) and immunotherapy response (26). To characterize hits from these and other genetic and genomic efforts, novel mouse strains need to be generated. However, creating mouse alleles and breeding multiallelic melanoma-prone experimental mice is expensive, slow, and cumbersome, rendering conventional mouse modeling an inefficient method to study gene functions in vivo. As an alternative, embryonic stem cell (ESC)-GEMMs (ESC-GEMM) have been previously developed (27). This approach relies on GEMM-derived ESCs that are modified in vitro and then used to generate chimeras by blastocyst injection. ESC-derived chimera tissues harbor the alleles of interest, thus enabling the use of chimeras as experimental mice (Fig. 1A). When combined with efficient ESC-targeting methods and advanced tools for modulating gene expression (28–30), ESC-GEMMs are a powerful strategy for rapid and versatile disease modeling, as has been demonstrated for breast, lung, and pancreatic cancer (31–34).

We report here the generation and validation of an ESC-GEMM platform for melanoma modeling. We derived ESC lines harboring 12 different combinations of driver alleles, as well as alleles to control the modulation of target gene expression. Our ESCs produce high-contribution chimeras that exhibit the same melanoma phenotypes as their conventionally bred counterparts. We tested applications of inducible genetic tools and employed them to assess the effects of Pten restoration in Pten-deficient melanomas. Moreover, we established melanoma cell lines that can complement ESC-GEMM experiments. Our platform is a powerful resource that will accelerate melanoma studies in vivo.

Twenty-eight-day-old females were superovulated by standard hormone injection procedure, paired with stud males, and blastocysts and late morulas were collected at 3.5 days post coitum. Embryos were plated in 2i medium and incubated until ESC colonies formed. ESC clones were expanded, genotyped (see Supplementary Materials and Methods for primer information), and the sex was determined. For ESC targeting, cells were nucleofected with targeting vector and pCAGGS-FLPe at a 2:1 ratio using the Lonza 4D-Nucleofector. ESCs were selected with hygromycin starting 48 hours following nucleofection and clones were picked after 7–10 days, expanded, and genotyped. For blastocyst injection, ESCs were feeder depleted, collected in M2 medium, and injected into Balb/c blastocysts by the Gene Targeting Core at Moffitt Cancer Center. Blastocysts were transferred into pseudo-pregnant CD-1 females. Detailed procedures are available in the Supplementary Materials and Methods.

All animal experiments were conducted in accordance with an Institutional Animal Care and Use Committee protocol (R-IS00003016) approved by the University of South Florida (Tampa, FL). All mouse strains used in this study were described previously: LSL-Braf^V600E (9), LSL-Nras^Q61R (10), Pten^Flox (14), Cdkn2a^Flox (15), Tyr-CreERt2 (13), CAGs-LSL-rtTA3 (29), and CHC (35). Chimerism was determined by the percentage of brown/black fur. Chimeras with at least 75% ESC contribution were used for almost all experiments.

4OHT (25 mg/mL, unless otherwise noted) dissolved in DMSO was administrated with a paintbrush on the shaved back skin of mice on 2 consecutive days at 3–4 weeks of age. Doxycycline feed was purchased from Envigo. Tumors were measured using calipers and volumes calculated using the formula (width² × length)/2.

Mouse tumors were dissociated using collagenase/dispase and trypsin. For recombination-mediated cassette exchange (RMCE), melanoma cells were cotransfected with targeting vector and pCAGGS-FLPe at a 2:1 ratio using FuGENE HD transfection reagent, and selected in hygromycin. Details can be found in the Supplementary Materials and Methods.

ESC and mouse melanoma cell lines were generated in the Karreth laboratory and were not authenticated. All cell lines were routinely Mycoplasma tested. In vitro assays, Western blotting, and IHC were performed using standard procedures. Experimental details and a list of antibodies can be found in the Supplementary Materials and Methods.

Statistical analysis was performed using GraphPad Prism software. Survival data were compared by applying the log rank (Mantel–Cox) test, and all other data were analyzed with the unpaired two-tailed t test or ordinary one-way ANOVA. A P value below 0.05 was considered statistically significant.

To alleviate the shortcomings of conventional melanoma GEMMs, we created an ESC-GEMM platform for versatile and rapid in vivo studies. We intercrossed mice carrying melanoma driver alleles and regulatory alleles to create 12 multiallelic strains harboring clinically relevant genotypes. We then derived ESC lines from each strain harboring combinations of Cre-inducible LSL-Braf^V600E (9) or LSL-Nras^Q61R (10) alleles and conditional knockouts of Pten (14) or Cdkn2a (Table 1; ref. 15). In addition, our ESCs harbor a melanocyte-specific, 4-Hydroxytamoxifen (4OHT)-inducible Cre recombinase allele (Tyr-CreERt2; ref. 13), a Cre-inducible Tet reverse transactivator (CAGs-LSL-rtTA3; ref. 29), and a homing cassette in the col1A1 locus (CHC; ref. 35) for efficient genomic integration of expression constructs via RMCE. This allele combination enables Cre-inducible recombination of driver alleles, as well as Cre- and doxycycline-inducible regulation of genes of interest in melanocytes (Fig. 1A). ESCs were derived and expanded in 2i medium (36), genotyped, and their sex determined. For all 12 genotypes, ESC lines having an undifferentiated morphology were selected to test their ability to produce chimeric mice. All tested ESC lines contributed to chimeras, and most ESCs produced multiple high-contribution chimeric mice having >75% ESC contribution (Fig. 1B).

We then tested whether the ESC-GEMM approach affects melanomagenesis by comparing the phenotypes of chimeras and conventionally bred mice. Three- to 4-week old chimeras of the various genotypes (see genotype abbreviations in Table 1) were topically treated with 4OHT on 2 consecutive days to induce Cre activity in melanocytes. Similarly, 4OHT was administered to conventionally bred LSL-Braf^V600E; Pten^Flox/Flox; Tyr-CreERt2 (BPP breeding), LSL-Braf^V600E; Cdkn2a^Flox/Flox; Tyr-CreERt2 (BCC breeding), LSL-Nras^Q61R; Pten^Flox/Flox; Tyr-CreERt2 (NPP breeding), and LSL- Nras^Q61R; Cdkn2a^Flox/Flox; Tyr-CreERt2 (NCC breeding) mice. Importantly, BPP, BCC, NPP, and NCC chimeras developed superficially growing tumors and displayed overall survival rates similar to their conventionally bred counterparts (Fig. 1C and D). Tumors from all genotypes were positive for melanoma markers MART1 and S100 (Supplementary Fig. S1A), and we observed no tumors in undesired tissues upon gross examination at necropsy. Similar to melanomas arising in conventionally bred mice, ESC-GEMM chimera tumors were predominantly amelanotic (Supplementary Fig. S1A). The various genotypes of our ESC-GEMM chimeras led to melanoma phenotypes with different penetrance and latency (Supplementary Fig. S1B and S1C), and tumor numbers (Supplementary Fig. S1D). All BPP chimeras and a considerable number of NPP chimeras developed too many tumors for individual tumor nodules to be discerned (denoted as >25 tumors in Supplementary Fig. S1D). When BPP chimeras were treated once with a 10-fold lower dose of 4OHT, fewer melanomas developed with the same latency (Supplementary Fig. S1E). The spectrum of melanoma phenotypes of the ESC-GEMM chimeras enables a wide range of applications on appropriate genetic backgrounds.

We next explored the efficiencies and shortcomings of inducible RNAi and CRISPR technologies developed for in vivo use (28, 30, 37) when combined with the ESC-GEMM approach. We first assessed whether gene depletion by the Cre/loxP technology, CRISPR-Cas9, or RNAi provokes similar melanoma phenotypes. We chose to target Pten on a Braf^V600E background as this combination results in an aggressive melanoma phenotype (16). We targeted B ESCs with either a Cre- and doxycycline-inducible Cas9 construct (38) that also contains a validated sgRNA against Pten (Fig. 2A; Supplementary Fig. S2A and S2B; ref. 39) or with a doxycycline-inducible, GFP-linked miR30-based RNAi construct containing a validated shRNA targeting Pten (shPten.1523; Fig. 2A; ref. 40). We also targeted B ESCs with a control sgRNA targeting a nongenic region in chromosome 8 (CR8; ref. 30) or a control shRNA against Renilla luciferase (shRen.713; ref. 41). For each targeting, we picked and expanded 6–12 clones. We confirmed integration of the targeting constructs into the CHC by PCR, which notably was successful in all tested clones.

Targeted ESCs maintained the ability to produce high-contribution chimeras (Supplementary Fig. S2C). We treated 3- to 4-week old B^{TRE-Cas9_sgPten} and B^TRE-shPten chimeras with 4OHT on their back skin to activate Cre in melanocytes. The B^{TRE-Cas9_sgPten} chimeras were fed doxycycline-containing chow for 14 days immediately following the 4OHT treatment to induce Cas9 expression. B^TRE-shPten chimeras were kept on a doxycycline diet (625 mg/kg) continuously to maintain expression of the shRNA. We compared melanoma development in the B^{TRE-Cas9_sgPten} and B^TRE-shPten chimeras to that in BP and BPP chimeras (generated with untargeted BP and BPP ESCs) where Cre deletes one or both copies of Pten, respectively. Melanomas in B^{TRE-Cas9_sgPten} chimeras developed with a slightly longer latency (Fig. 2B) and a markedly lower multiplicity (Fig. 2C) than in BPP chimeras. Surprisingly, B^TRE-shPten chimeras displayed increased melanoma latency and decreased tumor number compared with all other cohorts (Fig. 2B and C), which was accompanied by an extensive expansion of the cutaneous melanocyte population (Supplementary Fig. S2D and S2E). Control chimeras (B^{TRE-Cas9_sgCR8} and B^{TRE-shRen.713}) treated in a similar fashion did not develop any melanomas (Fig. 2C). Melanomas from B^{TRE-Cas9_sgPten}, B^TRE-shPten, BPP, and BP chimeras displayed reduced expression of Pten and increased phosphorylation of Akt (Supplementary Fig. S2F and S2G), demonstrating that while all three genetic methods of Pten depletion promote melanomagenesis, their efficiency and phenotype vary drastically.

To further test possible applications for CRISPR-Cas9 in ESC-GEMMs, we took advantage of the doxycycline-inducibility of our Cas9 construct (Fig. 2A). We first asked whether adjusting the duration of Cas9 expression could be used to regulate melanoma multiplicity. We fed B^{TRE-Cas9_sgPten} chimeras a doxycycline diet for 1, 3, or 7 days and monitored melanoma development. Similar to a colon cancer model (30), limiting the duration of Cas9 expression correlated with fewer melanomas (Fig. 2D; Supplementary Fig. S3A). Notably, the reduction in disease burden had no significant effect on survival (Supplementary Fig. S3B). Thus, adjusting the duration of Cas9 activity is suitable to control melanoma numbers without impacting the timing of tumor emergence and the rate of growth.

In the vast majority of compound-mutant GEMMs, all Cre-dependent alleles are induced simultaneously. We tested whether combining our ESC-GEMMs with inducible CRISPR-Cas9 enables sequential modeling of Braf^V600E expression and Pten depletion, which more closely mimics the sequence of events in human melanoma. To this end, we treated B^{TRE-Cas9_sgPten} chimeras with 4OHT to induce Braf^V600E expression followed by activating Cas9 expression 6 weeks later. Similar to simultaneous Pten deletion, delayed Pten deletion provoked a robust melanoma phenotype. Independent of the timepoint of Cas9 activation, the median chimera survival after Pten deletion was approximately 80 days, and both cohorts developed a comparable number of melanomas (Fig. 2E; Supplementary Fig. S3C and S3D), indicating that melanomagenesis is unaffected by delayed Pten deletion. These results demonstrate that inducible CRISPR-Cas9 can be utilized to model the sequential genetic events occurring in human melanomas.

Finally, conventionally bred LSL-Braf^V600E; Pten^FL/FL; Tyr-CreERt2 mice often develop spontaneous lesions due to leakiness of Tyr-CreERt2 (18). Undesired melanomas may require the censoring of experimental animals, resulting in a greater number of mice needed. Importantly, spontaneous melanomas did not occur in uninduced B^{TRE-Cas9_sgPten} chimeras (Supplementary Fig. S3E), indicating that Cre- and doxycycline-inducible CRISPR-Cas9 offers tighter spatiotemporal control over tumor suppressor deletion and melanoma induction.

The Pten-targeting shRNA potently silences Pten expression (40), and we confirmed highly efficient Pten silencing in vivo (Supplementary Fig. S2F). Thus, it was surprising that B^TRE-shPten chimeras survived significantly longer than BP chimeras, in which only one allele of Pten is inactivated (Fig. 2B). Adverse effects of doxycycline have been reported including on melanoma cells (42), and we tested whether doxycycline treatment contributed to the mild melanoma phenotype of B^TRE-shPten chimeras. We applied 1 μL aliquots of 4OHT to the back skins of conventionally bred LSL-Braf^V600E; Pten^FL/FL; Tyr-CreERt2 mice to induce focal melanomas. Immediately following 4OHT treatment, these mice were placed on one of two doxycycline diets (625 or 200 mg/kg) or kept on regular chow. Interestingly, mice given 625 mg/kg doxycycline chow developed fewer and smaller tumors compared with control and 200 mg/kg doxycycline-treated mice (Supplementary Fig. S4A and S4B), demonstrating that melanoma initiation and growth are impaired by doxycycline. Whether doxycycline has melanoma cell intrinsic effects or alters the tumor microenvironment, the immune system, or the microbiome remains to be determined. Nevertheless, this result may explain, at least in part, the phenotype of B^TRE-shPten chimeras.

In vivo gene silencing with doxycycline-inducible RNAi is reversible (28), enabling tumor suppressor restoration studies (43). To test the reversibility of inducible shRNAs and the effect of Pten restoration, melanoma-bearing B^TRE-shPten chimeras were taken off doxycycline. Doxycycline withdrawal diminished GFP fluorescence and expression in B^TRE-shPten melanomas (Fig. 3A–C), indicating inactivation of the shRNA construct, and resulted in Pten restoration (Fig. 3B and C) and reduced pAKT expression (Fig. 3B). Strikingly, Pten restoration completely prevented tumor growth (Fig. 3D; Supplementary Fig. S4C and S4D). The stalled tumor growth can be attributed to a decrease in proliferation, as shown by fewer Ki67-positive tumor cells (Fig. 3C and E). Thus, reversible gene silencing in melanoma ESC-GEMMs demonstrates the requirement for Pten loss for continued tumor growth.

We further examined Pten reexpression by a complementary approach using cDNA expression constructs. We targeted BPP ESCs with doxycycline-inducible cDNA constructs encoding Pten or GFP (TRE-Pten and TRE-GFP), created chimeric mice, and initiated melanoma development with 2.5 mg/mL 4OHT. We then placed BPP^TRE-Pten and BPP^TRE-GFP chimeras on 200 mg/kg doxycycline chow immediately following 4OHT administration to assess whether ectopic expression of Pten prevents melanoma development. While melanoma onset was similar between BPP^TRE-Pten and BPP^TRE-GFP chimeras (Fig. 4A), BPP^TRE-Pten chimeras exhibited increased overall survival (Fig. 4B) and developed fewer melanomas than BPP^TRE-GFP controls (Fig. 4C). IHC confirmed the expression of Pten and GFP in melanomas from BPP^TRE-Pten and BPP^TRE-GFP chimeras, respectively (Fig. 4D). However, only about half of the analyzed melanomas from doxycycline-treated BPP^TRE-Pten chimeras showed strong Pten expression while almost all tumors from BPP^TRE-GFP chimeras exhibited strong GFP staining (Fig. 4E; Supplementary Fig. S5A and S5B). Melanomas from BPP^TRE-Pten mice expressed rtTA3 (Supplementary Fig. 5C), indicating that while there is selection against expression of ectopic Pten, this is not because of failed recombination of the CAGs-LSL-rtTA3 allele. Thus, a complementary cDNA-based approach validates that Pten reactivation prevents melanomagenesis and highlights the versatility of our melanoma ESC-GEMM platform.

Melanoma cell lines established from GEMMs (44–46) can be used for syngeneic transplants in immunocompetent hosts and are invaluable resources supporting melanoma studies. We established melanoma cell lines (Supplementary Fig. S6A) from B^TRE-shPten and BPP chimeras to further complement our ESC-GEMM approach in general, and the Pten restoration experiments specifically. Our ESCs and thus melanoma cells derived from them are on an almost pure C57BL/6 background, enabling tumor formation upon transplantation in both immunocompromised Nu/Nu and immunocompetent C57BL/6 recipients (Supplementary Fig. S6B). We then tested the functionality of the regulatory alleles in BPP cells. We were able to insert a constitutively active EF1α-GFP construct into the CHC by RMCE in six of 11 cell lines (Supplementary Fig. S6C and S6D). In contrast, Cre was no longer expressed and delivery of a Cre reporter (47) together with 4OHT administration confirmed the absence of active Tyr-CreERt2 (Supplementary Fig. S6E and S6F). Finally, doxycycline administration induced GFP expression in cells infected with a doxycycline-inducible GFP lentivirus (TRE-GFP), validating the functionality of rtTA3 (Supplementary Fig. S6G and S6H).

We then withdrew doxycycline from B^TRE-shPten cells, which restored Pten expression, decreased Akt phosphorylation (Fig. 5A; Supplementary Fig. S7A), and impaired proliferation and focus formation (Fig. 5B and C; Supplementary Fig. S7B and S7C). Upon transplantation into Nu/Nu mice, tumors grew slower in the off doxycycline cohort compared with the on doxycycline group (Fig. 5D), resulting in increased survival (Fig. 5E). Interestingly, withdrawing doxycycline from mice bearing established tumors attenuated tumor growth (Fig. 5D), which also moderately increased survival (Fig. 5E). Next, we delivered lentiviral, doxycycline-inducible PtenWT, phosphatase-dead PtenC124S, or GFP cDNA constructs to BPP melanoma cells. Doxycycline treatment of cells harboring the PtenWT cDNA, but not the PtenC124S or GFP cDNAs, reduced pAkt (Fig. 5F; Supplementary Fig. S7D) and decreased proliferation and focus formation (Fig. 5G and H; Supplementary Fig. S7E and S7F). The tight regulation and induction of the cDNA constructs by doxycycline was verified in both mouse melanoma cell lines (Supplementary Fig. S7G). Finally, we transplanted PtenWT and PtenC124S BPP melanoma cells into C57BL/6 recipients on a doxycycline diet and observed slower growth of PtenWT tumors compared with PtenC124S tumors (Fig. 5I). In summary, melanoma cell lines isolated from our chimeras are a useful resource to complement ESC-GEMM experiments and study candidate gene functions in vitro and in vivo.

We have established a melanoma ESC-GEMM platform for rapid in vivo studies and demonstrate in this study its versatility and flexibility. Given that it takes less than 2.5 months from ESC targeting to inducing melanomagenesis in experimental chimeras, we anticipate that our platform has the potential to dramatically accelerate melanoma studies in mice. Importantly, both the ESC lines and chimera-derived melanoma cell lines will be available from our laboratory and distributed to the melanoma research community. When using the ESC-GEMM approach, several points should be considered. First, while our untargeted ESCs produce high-contribution chimeras, we found that lower contribution chimeras are more common when targeted ESC clones are used. Thus, targeted ESCs may produce surplus chimeras that need to be excluded from experiments due to their lower chimerism. As an alternative to injecting ESCs into blastocysts, tetraploid complementation or 8-cell embryos injection approaches could be used. While these techniques produce entirely or highly ESC-derived mice, respectively, their success rate is typically lower than regular blastocyst injection. Nevertheless, these techniques could be useful when exclusively high-contribution chimeras are needed.

Second, low-contribution chimeras may be useful for certain applications when highly aggressive backgrounds such as BPP are used. We have successfully used chimeras having 5%–10% contribution from BPP ESCs, where we induced focal tumors in the ESC-derived skin areas with small aliquots of 4OHT. This approach is practical when individual tumors are to be followed longitudinally. It remains to be determined whether such tumors originate from individual transformed melanocytes or if they are polyclonal. In polyclonal tumors, the Tet-ON system may not be operational in all tumor cells as recombination of the CAGs-LSL-rtTA3 allele may fail in tumor subclones. Thus, only a portion of the tumor cells may express or silence the gene of interest. This, in turn, could result in the emergence of “escaper” tumors in prevention or regression studies, where the gene of interest provokes strong negative selection. For such studies, it is advisable to use high-contribution chimeras and decrease the 4OHT concentration, which considerably reduces the number of melanomas on the BPP background. This increases the likelihood that each tumor originates from individual transformed melanocytes, thereby facilitating the mechanistic analysis of how putative escaper tumors arise. Thus, by varying the 4OHT concentration and means of application, high- and low-contribution chimeras may be used for different experimental contexts.

Third, 2i medium containing inhibitors against MEK and GSK3β is ideal for the derivation of ESC lines. However, prolonged culture in 2i medium may alter the epigenetic state of ESCs, thereby lowering their developmental potential (48). We maintained and targeted our ESCs in 2i medium, which may explain why our targeted ESCs produced more low-contribution chimeras. The use of alternative media, such as traditional serum- and LIF-containing medium or 2i with alternative inhibitors (48), may further improve the potential of our ESCs, but this has to be optimized for each ESC line individually.

We tested whether modulation of gene expression with different technologies, Cre/loxP, CRISPR, and RNAi, produces similar phenotypes. Notably, we found that melanoma latencies and tumor numbers were quite dissimilar. The long latency of melanomas induced by doxycycline-inducible shRNA was particularly surprising given the potency of shPten.1523 in vitro. While our findings of adverse effects of high doxycycline concentrations on melanoma initiation and growth provide an explanation for this observation, it remains to be determined whether there is a difference in silencing efficiency in tumors compared with cells in culture. Moreover, future studies are required to define the optimal doxycycline concentration for shRNA experiments where maximum silencing is achieved with minimal doxycycline effects, although this will likely be shRNA dependent. Nevertheless, we highlighted several advantages of doxycycline-inducible CRISPR-Cas9 and shRNA constructs in combination with the ESC-GEMM approach. While inducible CRISPR-Cas9 can be used to control melanoma multiplicity and the timing of tumor suppressor loss, inducible shRNAs enable reversible gene silencing. We used this approach to restore expression of Pten, which halted melanoma growth. We complemented this approach by inducibly expressing Pten cDNA in BPP chimeras, which restricted melanoma formation. Inhibition of the PI3K downstream effectors AKT or mTOR has limited efficacy in conventionally bred LSL-Braf^V600E; Pten^Flox/Flox; Tyr-CreERt2 mice, even in combination with a BRAF inhibitor (16). However, whether restoring PTEN functions in PTEN-deficient tumors constitutes a superior therapeutic approach remains to be investigated. Thus, our models will be useful for evaluating how the effects of Pten restoration compare with inhibition of downstream signaling pathways.

To complement ESC-GEMM experiments, we isolated murine melanoma cell lines from B^TRE-shPten and BPP chimeras. We show that Pten restoration lessens the aggressiveness of these cells in vitro, and diminishes tumor growth in vivo. Murine melanoma cell lines have recently been established and their ability to form tumors when transplanted into immunocompetent syngeneic recipients has made them an invaluable resource to the melanoma community (44–46). Similarly, our murine melanoma cell lines form tumors when transplanted into C57BL/6 recipients. Importantly, however, transgenic cassettes can be incorporated into the CHC via RMCE in BPP cells and their expression can be controlled using the integrated Tet-ON system. Thus, cancer gene function can be readily modulated and assessed in vitro and in vivo. Moreover, given the recent success of immunotherapy, we foresee immense utility of these cells for studies focused on delineating the molecular mechanisms governing melanoma immune suppression. In summary, we have generated and validated an ESC-GEMM platform consisting of a panel of newly derived ESCs and murine melanoma cell lines that will considerably facilitate future in vivo melanoma studies.

No potential conflicts of interest were disclosed.

Conception and design: I. Bok, F.A. Karreth

Development of methodology: O. Vera, F.A. Karreth

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Bok, X. Xu, N. Jasani, K. Nakamura, J. Reff, A. Nenci, J.G. Gonzalez, F.A. Karreth

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Bok, F.A. Karreth

Writing, review, and/or revision of the manuscript: I. Bok, O. Vera, J.G. Gonzalez, F.A. Karreth

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Nenci

Study supervision: F.A. Karreth

We are grateful to G. DeNicola for critical reading of the article. We thank D. Tuveson, N. Sharpless, S. Lowe, and M. Bosenberg for mouse alleles, and L. Dow, J. Zuber, and T. Jacks for plasmids. This work was supported by grants to F.A. Karreth from the NIH/NCI (K22 CA197058 and R03 CA227349), the Melanoma Research Alliance (MRA Young Investigator Award), the American Cancer Society (IRG-14-189-19), the Moffitt Skin SPORE (P50 CA168536) Career Development Program, a Miles for Moffitt Milestone Award, and a Harry J. Lloyd Charitable Trust Career Development grant. The Gene Targeting Core is supported in part by the NCI (P30 CA076292).