Conventional genetically engineered mouse models (GEMM) are time-consuming, laborious, and offer limited spatiotemporal control. Here, we describe the development of a streamlined platform for in vivo gene activation using CRISPR activation (CRISPRa) technology. Unlike conventional GEMMs, this model system allows for flexible, sustained, and timed activation of one or more target genes using single or pooled lentiviral guides. Myc and Yap1 were used as model oncogenes to demonstrate gene activation in primary pancreatic organoid cultures in vitro and enhanced tumorigenic potential in Myc-activated organoids when transplanted orthotopically in vivo. Implementation of this model as an autochthonous lung cancer model showed that transduction-mediated activation of Myc led to accelerated tumor progression and significantly reduced overall survival relative to nontargeted tumor controls. Furthermore, Myc activation led to the acquisition of an immune suppressive, “cold” tumor microenvironment. Cross-species validation of these results using publicly available RNA/DNA-seq datasets linked MYC to a previously described immunosuppressive molecular subtype in patient tumors, thus identifying a patient cohort that may benefit from combined MYC- and immune-targeted therapies. Overall, this work demonstrates how CRISPRa can be used for rapid functional validation of putative oncogenes and may allow for the identification and evaluation of potential metastatic and oncogenic drivers through competitive screening.

Significance:

A streamlined platform for programmable CRISPR gene activation enables rapid evaluation and functional validation of putative oncogenes in vivo.

Improved technologies for introducing controlled perturbations to gene expression in vivo hold great promise for the interrogation of gene function and modeling of human disease. While conventional genetically engineered mouse models (GEMM) require that a custom model be created for each individual gene-of-interest, a programmable mouse model for gene activation, where one or more genes can be activated in an ad hoc manner, would constitute a powerful tool for cost and time efficient modeling of genetic drivers in human disease, including cancer.

Since its description as a programmable system for gene editing (1–3), CRISPR/Cas9 has become a ubiquitous scientific tool for gene knockout (CRISPR KO), interference (CRISPRi), and activation (CRISPRa). Although systems for in vivo CRISPR KO have been implemented using mice conditionally expressing Cas9 (LSL-Cas9 mice; ref. 4) and RNA guides delivered through viral transduction (5) or plasmid electroporation (6, 7), the incorporation of CRISPRa in in vivo platforms has to date been limited. The CRISPR/Cas9 Synergistic Activation Mediator (SAM) system is a robust three-component system for CRISPRa, consisting of dCas9 fused to four tandem repeats of the herpes simplex transactivator VP16 (dCas9VP64), a transactivating fusion protein (MS2-p65-HSF1) and MS2 aptamer-loop modified single guide RNA [sgRNA(MS2); refs. 8, 9]. The SAM system has been used to drive expression of individual or multiple genes using single or pooled RNA guides across a wide range of cell types in vitro (8–11), and was recently implemented in vivo (12).

In contrast with CRISPR KO, CRISPRa relies on stable integration of the RNA guide into the host genome for retained guide expression, sustained gene activation, and tracing of guide abundance over long periods of time, for example, during tumor formation. Therefore, for oncogene studies, lentiviral delivery (13, 14) of sgRNA sequences is favorable over transient adeno-associated virus or adenovirus systems (15–17).

Unlike standard transgene-based overexpression technologies, CRISPRa relies on recruitment of transactivators to the endogenous gene locus, and thus more faithfully retains important aspects of transcriptional diversity and regulation, limits the level of overexpression, enables isotype-specific activation, and allows for noncoding transcripts and protein-coding genes that are too large for transduction/transfection to be activated (8).

In addition to in vivo gene activation, a CRISPRa-enabled mouse model would serve as an indefinite source of CRISPRa competent primary cells. In vivo CRISPRa oncogene activation elegantly takes advantage of the fact that a gene activation event that leads to increased tumorigenicity will be positively selected for, which creates the basis for a platform for in vivo oncogene evaluation and screening.

With this in mind, we created a model for oncogene activation consisting of a GEMM harboring conditional expression of the components necessary for CRISPRa/SAM, in combination with Cre/sgRNA-encoding lentivirus that can be introduced in a targeted manner through local administration. Although we hypothesize that a wide range of human tumor entities can be modeled using this approach, we describe a first proof-of-principle implementation of our system by targeting MYC and YAP1, two important oncogenes across human cancer, in pancreatic organoids and in an autochthonous lung cancer model.

Using our LSL-SAM mouse model, we show successful Myc and Yap1 gene activation in pancreas organoid cultures in vitro, and enhanced tumorigenicity following orthotopic transplantation of Myc-activated organoids. By incorporating conditional P53 loss and oncogenic Kras in our LSL-SAM platform to create an autochthonous lung cancer model, we triggered gene activation and tumor formation in vivo using nasal instillation of Cre/sgRNA lentivirus.

Moreover, Myc-activation resulted in MYC-associated transcriptomic reprogramming, accelerated tumor progression, and significantly reduced survival. Consistent with work of others (18–21), we found a pronounced cold immune microenvironment in Myc-activated tumors. We furthermore identified Myc as a driver of the immunosuppressive lung adenocarcinoma subtype 2 (LuAd2) in mouse and human lung tumors.

Using our in vivo CRISPRa/SAM platform, we were able to recapitulate the tumorigenic and immune evasive role of MYC in human lung adenocarcinoma. In addition, our work provides cross species evidence that MYC is a driver of the LuAd2 subtype (22), highlighting a previously unexplored opportunity for targeted therapy in this recalcitrant subtype.

Generating the LSL-SAM mouse

To target conditional dCas9VP64, MS2-p65-HSF1, and mCherry expression to the murine Rosa26 locus, we generated the pLSL-dCas9VP64-P2A-MS2-p65-HSF1-T2A-mCherry-Rosa26TV (LSL-SAM) vector: A 6.7 kbp fragment encoding for tricistronic expression of dCas9VP64, MS2-p65-HSF1 and mCherry was amplified from the PB-UniSAM vector and assembled into the BstBI/NsiI digested empty backbone of the pLSL-Cas9-Rosa26TV vector, using HiFi DNA assembly (NEB). The LSL-SAM insert from one bacterial colony was sequence-verified trough NGS de novo assembly (Applied Biological Materials Inc.), linearized (XhoI) and electroporated into G4 mES cells. Following selection with G418 and screening with PCR reactions spanning the left and right homology arms, founder mice were generated through blastocyst injection and implantation in B6 albino females. Electroporation, selection, colony picking, and blastocyst implantation was performed by the MD Anderson Cancer Center Genetically Engineered Mouse Facility (GEMF). LSL-SAM mice were genotyped using a 3-primer PCR reaction that recognizes the R26(LSL-SAM) construct as well as the wild-type R26 allele, see Supplementary Table S7 for genotyping primers.

Mouse models

All studies were conducted in compliance with the institutional guidelines [MD Anderson; Institutional Animal Care and Use Committee (IACUC)]. The following mouse models were used in these studies:

Rosa26LSL-SAM (LSL-SAM), Rosa26LSL-YFP (LSL-YFP, Rhim 2012), Ptf1aCre/+;Rosa26LSL-SAM/LSL-YFP (CSY), Ptf1aCre/+;Rosa26LSL-YFP/+ (CY), P53FlF;KrasLSL-G12D/+;Rosa26LSL-YFP (PPKY; ref. 23), P53FlF;KrasLSL-G12D/+;Rosa26LSL-SAM (PPKS). Mice were backcrossed to a C57BL/6NJ background (see Supplementary Table S7). All animals were archived, and mice with the same genotypes were assigned randomly to experimental groups. All experiments were performed on balanced cohorts of male and female mice. We found no sex-specific differences in disease initiation and progression. All mice with the appropriate genotype were included in the study. All mice received standard chow diet ad libitum and were housed in pathogen-free facility (modified barrier) with standard controlled conditions. No more than 5 mice were housed together, under the supervision of DVMS veterinarians in an Association for Assessment and Accreditation of Laboratory Animal Care–accredited animal facility at the University of Texas M.D. Anderson Cancer Center. The number of mice used for each experiment is listed as (n). All animal procedures were reviewed and approved by the MDACC IACUC (IACUC 00001626, PI: A.D. Rhim). Maximal tumor burden (10% of body weight) and maximal tumor size allowed by the ethics committee was not exceeded. No statistical methods were used to pre-determine sample sizes for in vivo experiments, but our sample sizes are similar to those reported in previous publications (24).

Cell lines

Lenti-X 293T cell line (Takara; #632180) and tumor-derived murine cell lines were cultured in DMEM with 10% FBS with 1x penicillin/streptomycin at 37°C in a humidified incubator with 5% CO2 and tested regularly for Mycoplasma (see Supplementary Table S8).

Copy-number assay

LSL-SAM copy number was determined using the TaqMan copy-number assay following the manufacturer's instructions using a custom mCherry TaqMan assay and Tfrc as the reference gene. Mice with known single mCherry copy number were used as controls.

Cre/CRISPRa backbone vector cloning

The pLV-Ef1a-Cre-U6-sgRNA(MS2) vector was generated by replacing the BleoR cassette with Cre recombinase in the plenti sgRNA(MS2)_zeo backbone using PCR and HiFi assembly (NEB). One BsmBI site in the Cre cassette was destroyed through targeted mutagenesis. The pLV-CMV-Cre-U6-sgRNA(MS2) vector was generated by replacing the Ef1a promoter with a CMV promoter cassette in pLV-Ef1a-Cre-U6-sgRNA(MS2) using PCR and HiFi assembly (NEB). The pLV-CMV-CreERT2-P2A-[puroR/zeoR/hygR]-U6-sgRNA(MS2) vectors were generated by replacing the Ef1a-puroR cassette in plenti_sgRNA(MS2)_puro with CMV-CreERT2-P2A-[puroR/zeoR/hygR] cassettes. Two BsmBI sites in the CreERT2 cassette were destroyed through targeted mutagenesis. All generated plasmids were sequence verified with Sanger sequencing (see Supplementary Table S9). All reagents are listed in Supplementary Table S10.

Cre/CRISPRa guide cloning

sgRNA sequences were cloned into the Cre-encoding lentiviral backbones with Golden Gate cloning using a previously described protocol (8). Briefly, 100 pmol of sgRNA containing DNA oligos (IDT) were phosphorylated with 5 U of T4 PNK for 30 minutes at 37°C, followed by heating to 95°C and cooling to 25°C at a rate of 0.1°C/s to anneal into duplexes.

Golden Gate reaction mixtures were assembled containing 10 fmol annealed duplex, 25 ng lentiviral backbone, 500 U T7 ligase and 10 U BsmBI-v2. The reaction mix was incubated at 37°C for 5 minutes followed by 5 minutes at 25°C for a total of 15 cycles and transformed into Stbl3 bacteria. Successful cloning was confirmed by Sanger sequencing (hU6-F sequencing primer, see Supplementary Table S11 for sgRNA and primers).

Large-scale lentivirus production and titer quantification

Lenti-X cells were seeded in antibiotic-free complete DMEM media (10% FBS, 1 mmol/L sodium pyruvate). After overnight incubation, the cells were treated with 25 μmol/L chloroquine diphosphate for 5 hours. psPAX2, pCMV-VSV-G and transfer plasmid were mixed at a molar ratio of 1:0.54:1.26 in Opti-MEM (Gibco), using a total of 10 pmol plasmid per 150-mm plate. Linear polyethylenimine (PEI, 25,000 Da) was mixed with OptiMEM at 3-μg PEI per μg DNA. The plasmid mixture was mixed with the PEI solution and allowed to incubate for 20 minutes at room temperature, followed by addition to the LentiX cells and overnight incubation. Lentiviral supernatants were collected after 48 hours. Lentivirus supernatants were clarified through centrifugation (7 minutes at 500 × g) and filtration (0.45 μmol/L PES). For lentiviral particle concentration, the supernatant was mixed with 8.5% (w/v) PEG6000 and 0.4 mol/L sodium chloride, and incubated for 2 hours at 4°C. Lentiviral particles were precipitated through centrifugation at 1,500 × g for 45 minutes at 4°C and resuspended in PBS at a nominal concentration factor of 100x.

To allow for lentivirus titer quantification, a color-switching Cre-reporter cell line (referred to as LRLG) was generated by transducing LentiX cells with CMV-LoxP-DsRed-LoxP-eGFP lentivirus at low MOI. To determine Cre lentivirus titer, LRLG cells were seeded in lentivirus-containing media, supplemented with 8-μg/mL polybrene. Following 48-hour incubation, the fraction of eGFP-positive cells was determined using an Accuri C6 flow cytometer, and original virus titer was back-calculated. Only wells with an eGFP-positive fraction of 10%–30% were used for calculations. All reagents are listed in Supplementary Table S10.

Pancreatic organoid establishment and culture

Normal mouse pancreas organoids were established, maintained, and propagated as previously described (25). Briefly, a roughly 5 × 5 mm piece of pancreas tissue was minced in Hanks' Balanced Salt Solution (HBSS) and incubated with pancreas digestion media (0.4 mg/mL Collagenase P in HBSS). Digested pancreas was seeded in 50 μL Matrigel domes. When solidified, Matrigel domes were covered with organoid media supplemented with 10.5 μmol/L ROCK inhibitor (Y-27632). ROCK inhibitor was added only when organoids were seeded or passaged. 0.25 μg/mL Amphotericin B was added in the initial passage to prevent fungal contamination. Complete mouse pancreas organoid media consisted of Advanced DMEM/F-12 with 1x GlutaMAX, 10 mmol/L HEPES and 1x primocin media, supplemented with 0.5 μmol/L TGF-beta inhibitor A83–01, 0.05 μg/mL mEGF, 0.1 μg/mL hFGF-10, 0.1 μg/mL mNoggin, 0.01 μmol/L gastrin I, 1.25 mmol/L N-acetylcysteine, 10 mmol/L nicotinamide, 1x B27 supplement, 50% v/v Wnt3a-conditioned media, and 10% v/v R-spondin I conditioned media. For flow cytometry, transplantation and transduction experiments Matrigel domes were disrupted in media containing 2 mg/mL dispase, followed by TrypLE digestion with intermittent mechanical disruption. DNaseI was added if clumping was observed. The cell pellet was then resuspended in FACS buffer (2% FBS, 2 mmol/L EDTA in PBS) for flow cytometry or serum-free DMEM for transplantation. mCherry+ expressing cells were sorted on the Aria Cell Sorter II (BD Biosciences). DAPI was used to eliminate dead cells. All reagents are listed in Supplementary Table S10.

Organoid transduction

Single-cell suspensions were mixed with lentiviral particles and polybrene (8 μg/mL) in ultra-low attachment plates. Following overnight incubation, supernatant was replaced with complete organoid media supplemented with 2% Matrigel. After 4 days of culture, puromycin or hygromycin, and 4-hydroxytamoxifen (4OHT) were added. After 10 days of selection, organoid domes were seeded as normal.

qPCR

Organoid RNA was isolated using TRizol and the miRNeasy kit following the manufacturer's instructions, using 0.5 mL TRizol per 50 μL Matrigel dome and including on-column DNase treatment. cDNA was generated using the SuperScript III First-Strand Synthesis System (Invitrogen) and qPCR was performed using Power SYBR Green PCR Master Mix and a QuantStudio3 thermocycler. Primer sequences are listed in Supplementary Table S11. Transcript abundance was calculated as the difference relative to the housekeeping gene Actb (2–ΔCt) and to the control sample (ΔΔCt method). QuantStudio and Excel software were used to analyze the data.

Immunofluorescence

Cultured cells

Cultured cells were fixed in 10% formalin, blocked with 10% normal donkey serum/1% BSA/0.1% saponin, stained for dCas9VP64, MYC, YAP1 and/or γH2AX, followed by conjugated secondary antibodies and DAPI.

Tissue sections

Pancreatic and lung tumor tissues were fixed in zinc/formalin, embedded in paraffin and microtome sectioned. Sections were subjected to citrate antigen retrieval for 15 minutes at 95°C, blocked with 10% normal donkey serum/2% BSA/0.3% Triton X-100, stained for dCas9VP64, MYC and/or YAP1 followed by conjugated secondary antibodies and DAPI (Supplementary Table S12), quenched using the TrueVIEW Autofluorescence Quenching Kit and mounted with ProLong Diamond.

Organoid wholemount IF

Organoids growing in Matrigel domes were fixed in ice-cold 1:1 DMSO:methanol, blocked with 10% normal donkey serum/2% BSA/0.1% saponin, stained for dCas9VP64, MYC, YAP1 followed by conjugated secondary antibodies and DAPI. All reagents are listed in Supplementary Table S10.

Epifluorescent and confocal imaging

Epifluorescent imaging was performed on an Olympus IX73 inverted fluorescent microscope. Confocal imaging was performed using an Olympus FV1000 laser scanning microscope.

Quantification of γH2AX foci

Immunofluorescence-labeled cells cultured in chamber slides were scanned using a Cytation 5 Imaging Multi-Mode Reader (BioTek). Cell nuclei were identified using DAPI and the abundance of γH2AX foci was determined using CellProfiler (v.4.1.3).

MTT drug response assay

Cellular drug response was determined as previously described (26), briefly, 5,000 cells were seeded per 96-well and cells were treated with varying concentrations of oxaliplatin for 96 hours. MTT was added (0.5 mg/mL) for 4 hours, supernatant removed, and formazan crystals dissolved in acidified isopropanol (50 mmol/L HCl, 0.1% Triton X-100). Absorbance was read at 560 nm with 690 nm as reference, on a Cytation 3 Imaging Multi-Mode Reader (BioTek). Oxaliplatin IC50 values were calculated using curve-fitting in Prism (v.9.0.0). All reagents are listed in Supplementary Table S10.

Immunoblot analysis

Primary lung tumor cells and organoid cell lines were processed for immunoblotting according to a previously described protocol (24). Antibodies are listed in Supplementary Table S12. Blots were analyzed using a ChemiDoc XRS (Bio-Rad). For pharmacological MYC suppression, cells were seeded in 6-well plates and incubated with varying concentrations of MYCi361 for 48 hours followed by immunoblot analysis for MYC and β-actin. All reagents are listed in Supplementary Table S10.

Quantitative IHC and histology

IHC and hematoxylin and eosin (H&E) staining were performed as previously described (27). Tissue sections (3–5 μm) were stained with H&E or used for IHC, using standard avidin–biotin histologic methods. Antibodies are listed in Supplementary Table S12. Quantification was done using Aperio ImageScope (v.12.4.3.5008). All reagents are listed in Supplementary Table S10.

Intranasal instillation and orthotopic pancreatic organoid transplantations

All procedures were performed in a vaporizer applying isoflurane 2% and O2 flow 2 L/min. All mice were between 8 and 12 weeks of age. Intranasal instillations have been described previously (17). In brief, mice were anesthetized (to retain deep inhalation), placed on their back and the virus solution (80 μL) was administered dropwise and slowly into one nostril until virus was completely inhaled. Lung tissues were collected, fixed, embedded in paraffin and stained with H&E, for YFP, MYC, YAP1, dCas9VP64, and immune cell markers (CD45, CD8, CD4, CD19, and F4/80), proliferation (ki67), and apoptosis marker (cleaved caspase-3) as described above. For dose-dependent transfection efficiency and to confirm long-term stable transfection GFP/YFP/Cre-positive cells in pancreatic tissues 10 and 28 days after virus delivery were quantified using APERIO/image scope. For survival time point, tumor incidence and tumor size were quantified and representative macroscopic and microscopic H&E and YFP/mCherry pictures were taken. Orthotopic transplantations of pancreatic organoids were performed as previously described (25). In brief, 5×105 organoid cells prepared as single cell solutions as described above were injected into the pancreas of C57BL/6NJ PPKS syngeneic littermates using a 26 G needle. The mice were sacrificed at survival timepoint, defined when the mice reached sickness criteria. Tumor incidence and tumor sizes were quantified in a blinded manner and representative macroscopic (and mCherry+) and microscopic images were taken.

Tumor mRNA-seq and expression profiling

Total RNA was isolated from 16 mouse lung tumors (4 PPKY/LV, 4 PPKS/NT, 4 PPKS/M and 4 PPKS/Y) using the AllPrep DNA/RNA Mini Kit (Qiagen). Indexed libraries were generated using the KAPA mRNA HyperPrep kit and sequenced on a NextSeq500 sequencer (NextSeq 500/550 High Output Kit v2.5, 75 SE). Adapter trimming and FASTQ generation was done with Basespace (Illumina). Reads were aligned to mouse reference genome (mm10) and differential expression was determined using the RNA-Seq Alignment (v.2.0.2) and RNA-Seq Differential Expression (v1.0.1) Basespace applications (Illumina). Gene Set Enrichment Analysis (GSEA) was performed using the WebGestalt online portal (http://www.webgestalt.org/) and GSEAJava (v.4.1.0). Genes with an adjusted P value (q value) less than 0.05 were considered significant. Data visualization was performed in R (v.3.6.0) using Rstudio (v.1.1.463). Heatmaps were generated using heatmap.2 (gplots, v.3.6.3). GSEA dot plots were generated with ggplot2 (v.3.3.3). Raw data were deposited in ArrayExpress, accession E-MTAB-11122.

Mouse tumor transcriptomic subtyping

Mouse lung tumors were subtyped using CMScaller (v.0.9.2; ref. 28). Lists of subtype-specific genes were downloaded from Jang and colleagues (22). Orthologs for overexpressed genes from each subtype were generated using DIOPT (v.8.5, https://www.flyrnai.org/cgi-bin/DRSC_orthologs.pl), and can be found in Supplementary Table S6. Only upregulated genes from each subtype were used for CMScaller subtyping. Gene symbol, entrez id and expression value (rlog) for each sample was used as CMScaller input.

The Cancer Genome Atlas bioinformatics

Gene expression and copy-number data from the The Cancer Genome Atlas (TCGA)-LUAD study were downloaded from Xena (http://xena.ucsc.edu/). TCGA-LUAD tumors were subtyped using CMScaller (v.0.9.2; ref. 28). Lists of subtype-specific overexpressed genes were downloaded from Jang and colleagues (Supplementary Table S6; ref. 22). Gene symbol, entrez id and expression value (fpkm-uq) for each sample was used as CMScaller input. Differential Expression was performed in R using DESeq2 (v.1.26.0). GSEA for MYC-associated gene sets was performed using GSEAJava (v.4.1.0). Data visualization was performed in R (v.3.6.0) using Rstudio (v.1.1.463). Heatmaps were generated using heatmap.2 (gplots, v.3.6.3).

Schematics

Schematics were generated using BioRender (https://biorender.com/).

Data availability

Raw lung tumor mRNA-seq data were deposited in ArrayExpress, accession E-MTAB-11122.

Code availability

The codes supporting the current study are available from the corresponding author on request.

Materials availability

Plasmids generated in this study are available on request.

Quantification and statistical analysis

Description of sample size (n) and statistical tests are detailed in the figure legends or methods. All values for n are for individual mice or individual sample (biological replicates) unless otherwise noted. Sample sizes were chosen on the basis of previous experience with given experiments. Data are expressed as mean ± SEM unless otherwise noted. Differences were analyzed by an unpaired two-tailed Student t test, or by one-way ANOVA with Tukey multiple comparison test. The log-rank (Mantel–Cox) test was applied to determine survival significance and the Pearson r and two-sided test calculated for correlations. Statistical analysis was performed with Prism (GraphPad, v.9.0.0). P values ≤ 0.05 were considered significant. *, 0.01 < P ≤ 0.05; **, 0.001 < P ≤ 0.01; ***, 0.0001 < P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Development of a programmable mouse model for in vivo gene activation

To generate a mouse model for programmable and traceable gene activation in vivo, we created the LSL-SAM mouse by knocking in conditional, CAG-driven, loxp-stop-loxp (LSL) controlled, tricistronic expression of dCas9VP64, MS2-p65-HSF1, and mCherry in the Rosa26 (R26) mouse locus (Fig. 1AC; Supplementary Fig. S1A–S1B). To confirm functional activity of the construct, we transduced pancreatic ductal organoids from R26(LSL-SAM) and YFP reporter [R26(LSL-YFP)] mice with Cre adenovirus (AdCre) in vitro, resulting in LSL recombination and reporter (mCherry/YFP) expression (Supplementary Fig. S1C–S1D). To confirm conditional transgene expression in vivo, we generated CSY (Ptf1a(Cre/+) R26(LSL-SAM/LSL-YFP) mice, which were found to co-express mCherry and YFP (Fig. 1D) and nuclear dCas9VP64 in the pancreas (Fig. 1E and F). To test our ability to trigger expression of the SAM components in vivo, we delivered Cre lentivirus directly to the pancreas via retrograde ductal injections in R26(LSL-SAM) mice, resulting in sporadic expression of nuclear dCas9VP64 (Fig. 1G).

Figure 1.

Development and validation of the LSL-SAM mouse model. A, Schematic of the CRISPR activation/SAM platform. B, PCR of R26 knockin (right and left homology arms), intact LSL cassette, and LSL-SAM genotyping. C, LSL-SAM copy-number determination, one-way ANOVA with Tukey multiple comparison test. D, Fluorescent pancreatic acinar clusters from CSY and CY mice. Scale bar, 200 μm. E, Immunoblot of dCas9VP64 and YFP in pancreatic tissues from CSY and CY mice and total protein loading control. F, Immunofluorescence for dCas9VP64 in pancreas tissues from CSY and CY mice, dCas9VP64 (green), α-amylase (red), DAPI (blue). Scale bar, 20 μm. G, Sporadic nuclear expression of dCas9VP64 in pancreas tissues from R26(LSL-SAM/+) mice transduced with CMV-Cre lentivirus 10 days after transduction and untreated control, dCas9VP64 (green), DAPI (blue). Scale bar, 20 μm. ****, P ≤ 0.0001; ns, nonsignificant.

Figure 1.

Development and validation of the LSL-SAM mouse model. A, Schematic of the CRISPR activation/SAM platform. B, PCR of R26 knockin (right and left homology arms), intact LSL cassette, and LSL-SAM genotyping. C, LSL-SAM copy-number determination, one-way ANOVA with Tukey multiple comparison test. D, Fluorescent pancreatic acinar clusters from CSY and CY mice. Scale bar, 200 μm. E, Immunoblot of dCas9VP64 and YFP in pancreatic tissues from CSY and CY mice and total protein loading control. F, Immunofluorescence for dCas9VP64 in pancreas tissues from CSY and CY mice, dCas9VP64 (green), α-amylase (red), DAPI (blue). Scale bar, 20 μm. G, Sporadic nuclear expression of dCas9VP64 in pancreas tissues from R26(LSL-SAM/+) mice transduced with CMV-Cre lentivirus 10 days after transduction and untreated control, dCas9VP64 (green), DAPI (blue). Scale bar, 20 μm. ****, P ≤ 0.0001; ns, nonsignificant.

Close modal

Functional validation of the CRISPRa/SAM mouse model in organoid transplantation models

Next, to allow for programmable gene activation, we cloned lentiviral transfer plasmids encoding for Cre or tamoxifen-inducible Cre (CreERT2), and an antibiotic selection marker (puroR, zeoR or hygroR), in addition to human U6 (hU6) promoter-driven sgRNA(MS2; Fig. 2A). We next transduced and selected pancreatic organoids established from R26(LSL-SAM) mice with CreERT2/puroR lentivirus and administered 4OHT to trigger Cre recombination, resulting in SAM construct activation (Fig. 2B and C). To validate programmable gene activation, we transduced LSL-SAM organoids with lenti Cre/sgRNA targeting mouse Myc or Yap1, and a nontargeting control (see methods for guide sequences), which resulted in 4.4- and 2.8-fold transcriptomic activation, respectively, and protein-level overexpression (Fig. 2D and F).

Figure 2.

Exvivo gene activation in pancreatic organoids enhanced tumorigenic potential following Myc-activation. A, Top, schematic of the CMV-Cre-U6-sgRNA(MS2) and CMV-CreERT2-P2A-puroR/zeoR/hygR-U6-sgRNA(MS2) lentiviral vectors. Bottom, ex vivo gene activation in LSL-SAM pancreatic organoids. B, SAM construct activation pancreatic organoids. Red, mCherry. Scale bar, 100 μm. C, qPCR of dCas9VP64 in induced pancreas organoids (4OHT+), relative to controls (4OHT). D, qPCR in organoids transduced with Myc-targeted (sgMyc) and nontargeted (sgNT) lentivirus, n = 3. E, qPCR in LSL-SAM pancreatic organoids transduced with Yap1-targeted (sgYap) and nontargeted (sgNT) lentivirus, n = 3. F, Whole-mount IF for MYC (left) and YAP1 (right) in sgMyc, sgYap, and sgNT LSL-SAM pancreatic organoids. Green, MYC/YAP1; blue, DAPI. Scale bar, 100 μm. G, Whole-mount IF for MYC in PPKS/M and PPKS/NT pancreatic organoids. Green, MYC; blue, DAPI. Scale bar, 100 μm. H, qPCR for Myc in PPKS/M and PPKS/NT pancreatic organoids, n = 3. I, Schematic of orthotopic transplantation of oncogene-activated PPKS pancreatic organoids. J, Organoid-derived Myc-activated (TX PPKS/M) and nontargeted (TX PPKS/NT) pancreas tumors. Left, color image; right, macroscopic mCherry fluorescence image. Scale bar, 1 cm. K, H&E staining of TX PPKS/M and TX PPKS/NT tumors. Scale bars, 100 (left) and 20 μm (right). L, Immunofluorescence of MYC in TX PPKS/M and TX PPKS/NT tumors. Red, MYC; blue, DAPI. Scale bar, 20 μm. M, Immunoblot of dCas9VP64, MYC, and β-actin in TX PPKS/M and TX PPKS/NT tumors. N, Kaplan–Meier survival graph of TX PPKS/M and TX PPKS/NT-transplanted mice. Mantel–Cox (log-rank) test. C, D, E, and H, Significance tested with the unpaired two-tailed Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ****, P ≤ 0.0001.

Figure 2.

Exvivo gene activation in pancreatic organoids enhanced tumorigenic potential following Myc-activation. A, Top, schematic of the CMV-Cre-U6-sgRNA(MS2) and CMV-CreERT2-P2A-puroR/zeoR/hygR-U6-sgRNA(MS2) lentiviral vectors. Bottom, ex vivo gene activation in LSL-SAM pancreatic organoids. B, SAM construct activation pancreatic organoids. Red, mCherry. Scale bar, 100 μm. C, qPCR of dCas9VP64 in induced pancreas organoids (4OHT+), relative to controls (4OHT). D, qPCR in organoids transduced with Myc-targeted (sgMyc) and nontargeted (sgNT) lentivirus, n = 3. E, qPCR in LSL-SAM pancreatic organoids transduced with Yap1-targeted (sgYap) and nontargeted (sgNT) lentivirus, n = 3. F, Whole-mount IF for MYC (left) and YAP1 (right) in sgMyc, sgYap, and sgNT LSL-SAM pancreatic organoids. Green, MYC/YAP1; blue, DAPI. Scale bar, 100 μm. G, Whole-mount IF for MYC in PPKS/M and PPKS/NT pancreatic organoids. Green, MYC; blue, DAPI. Scale bar, 100 μm. H, qPCR for Myc in PPKS/M and PPKS/NT pancreatic organoids, n = 3. I, Schematic of orthotopic transplantation of oncogene-activated PPKS pancreatic organoids. J, Organoid-derived Myc-activated (TX PPKS/M) and nontargeted (TX PPKS/NT) pancreas tumors. Left, color image; right, macroscopic mCherry fluorescence image. Scale bar, 1 cm. K, H&E staining of TX PPKS/M and TX PPKS/NT tumors. Scale bars, 100 (left) and 20 μm (right). L, Immunofluorescence of MYC in TX PPKS/M and TX PPKS/NT tumors. Red, MYC; blue, DAPI. Scale bar, 20 μm. M, Immunoblot of dCas9VP64, MYC, and β-actin in TX PPKS/M and TX PPKS/NT tumors. N, Kaplan–Meier survival graph of TX PPKS/M and TX PPKS/NT-transplanted mice. Mantel–Cox (log-rank) test. C, D, E, and H, Significance tested with the unpaired two-tailed Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ****, P ≤ 0.0001.

Close modal

Myc activation in pancreatic organoids accelerates tumor formation

To test functional gene activation in the context of oncogenic drivers, we triggered Myc activation in “pre-oncogenic” PPKS [P53(F/F) Kras(LSL-G12D/+) R26(LSL-SAM)] organoids, resulting in 3.7-fold Myc transcriptional activation (Fig. 2G and H; Supplementary Fig. S2A–S2B). Yap1 activation yielded similar results (Supplementary Fig. S2C–S2D). To test whether Myc activation in the context of P53 loss and oncogenic Kras increases organoid tumorigenicity, we transplanted Myc-activated (PPKS/M) and nontargeted (PPKS/NT) ductal pancreatic organoids orthotopically into the pancreata of syngeneic mice (Fig. 2I). Transplanted organoids initiated tumors histologically consistent with pancreatic ductal adenocarcinoma (Fig. 2J and K). Myc-activated organoids generated MYC-overexpressing tumors (Fig. 2L and M) and significantly accelerated tumor progression in transplanted mice (median survival 92 vs. 133 days, P = 0.001, Fig. 2N; Supplementary Table S1).

In vivo transduction-mediated tumor initiation in an autochthonous lung cancer model

To allow for programmable gene activation in vivo, we developed a transduction approach using Cre/sgRNA encoding lentivirus. We generated a color-switching Cre-reporter cell line and a protocol for functional Cre-lentivirus quantification (Fig. 3A and B, see Materials and Methods). To achieve sufficiently high lentiviral titers for in vivo use, lentiviral supernatants were concentrated using PEG/NaCl precipitation, routinely achieving functional titers of approximately 3e9TU/mL (Fig. 3C and D). Next, we used nasal instillation to administer lentivirus directly to the mouse lung (Fig. 3E). To show stable genomic incorporation and transgene expression, we treated mice with CMV-eGFP encoding lentivirus and found eGFP-expressing cells in the lung parenchyma after 10 days and 4 weeks (Fig. 3F). We determined the dose–response relationship of Cre lentivirus dose and resulting recombination in the lung. CMV-Cre lentivirus was administrated to YFP reporter mice and the fraction of YFP-expressing cells in the lung was determined after 10 days. We found a sigmoidal-like dose–response relationship between the viral dose and the resulting number of YFP-expressing cells in the lung parenchyma (Fig. 3G and H; Supplementary Table S2). Minimal transduction was observed at doses below 1e7 TU, whereas between 1e7 and 1e8 TU the observed transduction rate increased in proportion to the dose to reach a maximum of approximately 90 cells/mm2 at 1e8 TU. Increasing the dose beyond 1e8 TU did not increase the transduction rate. In addition, we compared the performance of CMV and Ef1a promoter-driven Cre and found that they performed equivalently (Fig. 3I; Supplementary Table S2). We next generated lung tumors by transducing PPKY [P53(F/F) Kras(LSL-G12D/+) R26(LSL-YFP)] mice with CMV-Cre lentivirus (1e8TU). The resulting tumors (referred to as PPKY/LV) were found to be positive for Cre recombination, to harbor stable genomic integration of the viral payload (Fig. 3J; Supplementary Fig. S3A–S3C), and Cre expression was detected in normal lung tissues 10 days following transduction and in transduction-initiated tumors (Fig. 3K).

Figure 3.

Optimization of lentiviral production, quantification, and transduction for in vivo gene activation. A, DsRed to GFP color-switching Cre-reporter cell line (LRLG). Scale bar, 100 μm. B, Lenti Cretiter quantification. Untransduced (left) and CMV-Cre–transduced LRLG cells (right). C, Comparison of PEG/NaCl-precipitation and ultracentrifugation (UC) of Cre lentivirus. D, Precipitation and ultracentrifugation concentration factor. E, Schematic of nasal instillation of Cre lentivirus in R26(LSL-YFP) mice. F, IHC for GFP in mouse lung parenchyma 10 days (middle) and 4 weeks (right) following CMV-GFP lentivirus treatment, and control (left). Scale bars, 100 μm. G, LSL-YFP recombination in R26(LSL-YFP) reporter mouse lung parenchyma 10 days following CMV-Cre lentivirus treatment. Scale bar, 100 μm. H, qIHC of YFP 10 days following CMV-Cre lentivirus treatment, n = 4–6 per concentration. I, CMV-Cre and Ef1a-Cre–driven YFP recombination. Unpaired two-tailed Student t test. J, PCR showing retained viral payload (Cre sequence, left), YFP (middle), and Kras (right) recombination in PPKY/LV tumors; somatic (tail) DNA as control. K, IHC of Cre in lung parenchymal 10 days after CMV-Cre lentivirus transduction (left) and in PPKY/LV tumors (right). Scale bar, 100 μm. C and D, Significance tested with one-way ANOVA with the Tukey multiple comparison test. ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Figure 3.

Optimization of lentiviral production, quantification, and transduction for in vivo gene activation. A, DsRed to GFP color-switching Cre-reporter cell line (LRLG). Scale bar, 100 μm. B, Lenti Cretiter quantification. Untransduced (left) and CMV-Cre–transduced LRLG cells (right). C, Comparison of PEG/NaCl-precipitation and ultracentrifugation (UC) of Cre lentivirus. D, Precipitation and ultracentrifugation concentration factor. E, Schematic of nasal instillation of Cre lentivirus in R26(LSL-YFP) mice. F, IHC for GFP in mouse lung parenchyma 10 days (middle) and 4 weeks (right) following CMV-GFP lentivirus treatment, and control (left). Scale bars, 100 μm. G, LSL-YFP recombination in R26(LSL-YFP) reporter mouse lung parenchyma 10 days following CMV-Cre lentivirus treatment. Scale bar, 100 μm. H, qIHC of YFP 10 days following CMV-Cre lentivirus treatment, n = 4–6 per concentration. I, CMV-Cre and Ef1a-Cre–driven YFP recombination. Unpaired two-tailed Student t test. J, PCR showing retained viral payload (Cre sequence, left), YFP (middle), and Kras (right) recombination in PPKY/LV tumors; somatic (tail) DNA as control. K, IHC of Cre in lung parenchymal 10 days after CMV-Cre lentivirus transduction (left) and in PPKY/LV tumors (right). Scale bar, 100 μm. C and D, Significance tested with one-way ANOVA with the Tukey multiple comparison test. ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Close modal

In summary, following successful high titer lentivirus production, reliable titer quantification, critical comparison of promoter and dose dependencies as well as the confirmation of persistent transgene expression, we were able generate lung tumors with stable genomic integration.

Oncogene activation accelerates transduction-initiated lung tumor progression

We next set out to test CRISPRa-mediated oncogene activation and tumor formation in vivo. We administered 1e8TU Cre/sgMyc or Cre/sgYap1 lentivirus through nasal instillation to PPKS [P53(F/F) Kras(LSL-G12D/+) R26(LSL-SAM/+)] mice, referred to as PPKS/M and PPKS/Y, respectively (Fig. 4A). In addition, we administered lentivirus encoding for Cre and a nontargeting guide to PPKS (PPKS/NT) and PPKY (PPKY/LV) mice to serve as nontargeted CRISPRa and non-CRISPRa controls, respectively (Supplementary Table S3).

Figure 4.

Tumor initiation and in vivo oncogene activation through lentiviral transduction in the CRISPRa/SAM lung tumor model. A, Schematic of the autochthonous CRISPRa/SAM gene activation lung tumor model. B, Macroscopic and fluorescent images of PPKY/LV (top), and PPKS/M lung tumors (bottom). Scale bar, 1 cm. C, Immunofluorescence of PPKS/M and PPKS/Y lung tumors, with PPKY/LV and PPKS/NT controls. Green, dCas9VP64; red, MYC/YAP1; blue, DAPI. Scale bar, 40 μm. D,Myc transcript expression (rlog) in PPKY/LV, PPKS/NT, PPKS/M, and PPKS/Y tumors. E,Yap1 transcript expression (rlog). F, Immunoblot of MYC, YAP1, and β-actin in PPKS/M, PPKS/Y, and PPKS/NT control lung tumors. G, Immunoblot of MYC and β-actin in MYCi361-treated PPKS/M tumor-derived cells. H, Kaplan–Meier survival analysis of PPKS/M, PPKS/Y, PPKY/LV, and PPKS/NT lung tumor mice. Pairwise Mantel–Cox (log rank) test. I, Representative tumor histology (H&E), ki67, and cleaved caspase-3 (CC3) staining in PPKS/M, PPKS/Y, PPKY/LV, and PPKS/NT lung tumors. Scale bar, 100 μm. J, Relative tumor growth for each cohort. K, Quantification of relative lung tumor area. L, qIHC of ki67 in lung tumors. M, qIHC of cleaved caspase-3 in lung tumors. D, E, and J, Significance tested with one-way ANOVA with the Tukey multiple comparison test. K–M, Significance tested with the unpaired two-tailed Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Figure 4.

Tumor initiation and in vivo oncogene activation through lentiviral transduction in the CRISPRa/SAM lung tumor model. A, Schematic of the autochthonous CRISPRa/SAM gene activation lung tumor model. B, Macroscopic and fluorescent images of PPKY/LV (top), and PPKS/M lung tumors (bottom). Scale bar, 1 cm. C, Immunofluorescence of PPKS/M and PPKS/Y lung tumors, with PPKY/LV and PPKS/NT controls. Green, dCas9VP64; red, MYC/YAP1; blue, DAPI. Scale bar, 40 μm. D,Myc transcript expression (rlog) in PPKY/LV, PPKS/NT, PPKS/M, and PPKS/Y tumors. E,Yap1 transcript expression (rlog). F, Immunoblot of MYC, YAP1, and β-actin in PPKS/M, PPKS/Y, and PPKS/NT control lung tumors. G, Immunoblot of MYC and β-actin in MYCi361-treated PPKS/M tumor-derived cells. H, Kaplan–Meier survival analysis of PPKS/M, PPKS/Y, PPKY/LV, and PPKS/NT lung tumor mice. Pairwise Mantel–Cox (log rank) test. I, Representative tumor histology (H&E), ki67, and cleaved caspase-3 (CC3) staining in PPKS/M, PPKS/Y, PPKY/LV, and PPKS/NT lung tumors. Scale bar, 100 μm. J, Relative tumor growth for each cohort. K, Quantification of relative lung tumor area. L, qIHC of ki67 in lung tumors. M, qIHC of cleaved caspase-3 in lung tumors. D, E, and J, Significance tested with one-way ANOVA with the Tukey multiple comparison test. K–M, Significance tested with the unpaired two-tailed Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Close modal

All mice treated with Cre lentivirus developed mCherry or YFP-positive lung tumors (Fig. 4B). To determine whether oncogene activation led to persistent Myc and Yap1 overexpression, we stained tumor sections for MYC, YAP1, and dCas9VP64, revealing marked nuclear MYC expression in PPKS/M tumors and predominantly cytoplasmic YAP1 expression in PPKS/Y tumors, relative to controls (Fig. 4C; Supplementary Fig. S4A–S4B). We also performed mRNA-seq on tumor tissues, which revealed 2.8- and 6.5-fold Myc overexpression in Myc-activated (PPKS/M) tumors, relative to non-targeted (PPKS/NT) and non-CRISPRa (PPKY/LV) tumors, respectively (Fig. 4D). Yap1 was significantly overexpressed (3.3-fold) in Yap1-activated (PPKS/Y) tumors relative to non-CRISPRa (PPKY/LV) tumors. We found 2.2-fold overexpression of Yap1 relative to non-targeted tumors (PPKS/NT), but the result did not reach statistical significance (Fig. 4E). To confirm protein-level overexpression, we performed immunoblotting, which revealed MYC and YAP1 overexpression in PPKS/M and PPKS/Y tumors, respectively (Fig. 4F). In addition, we confirmed sustained MYC overexpression in cell lines derived from PPKS/M tumors, relative to PPKS/NT controls, and retained dCas9VP64 expression in all PPKS cell lines tested (Supplementary Fig. S4C–S4E). Using a recently described MYC-inhibitor (MYCi361; ref. 21), we were able to pharmacologically suppress MYC in these cell lines (Fig. 4G).

Strikingly, Myc activation resulted in significantly decreased overall survival (median survival: PPKS/M 102 days, PPKS/NT 152.5 days, PPKY/LV 133 days, Fig. 4H; Supplementary Table S3), increased tumor burden (Fig. 4IK), and higher proliferation rates (Fig. 4I and L) relative to controls, whereas apoptosis remained unaffected (Fig. 4I and M). Although an increase in proliferation was observed in PPKS/Y relative to controls (Fig. 4L), no significant difference in tumor progression or survival was obvious between PPKS/Y and PPKS/NT or PPKY/LV controls, nor between PPKS/NT and PPKY/LV (Fig. 4H), indicating a specific oncogenic effect of Myc activation in the lung.

Myc activation drives an aggressive, immunosuppressed tumor phenotype

As Myc activation resulted in markedly accelerated tumor progression, increased tumor burden, and tumor cell proliferation, we analyzed the transcriptomic profile and phenotype of these tumors in more detail.

We analyzed the expression of the main Myc family members (Myc, Mycn, and Mycl) and the proximal MYC signaling network (Supplementary Table S4; ref. 29) in Myc-activated (PPKS/M) relative to non-targeted control (PPKS/NT) tumors. Myc-activated tumors displayed significant overexpression of Mxi1, Mlx, and Mxd3 in addition to Myc, whereas Mycn, Mycl, Mxd1, Mlxip, Mnt and Mxd4 were significantly downregulated, indicating a pronounced effect on MYC signaling (Fig. 5A). Downregulation of Mycn and Mycl in Myc-activated tumors may represent a compensatory mechanism in response to Myc-overexpression.

Figure 5.

Transcriptomic reprogramming in Myc-activated lung tumors from the CRISPRa/SAM model. A, Heatmap of differentially expressed MYC network members in PPKS/M relative to PPKS/NT tumors, Myc/Mycn/Mycl (red). B, Heatmap of 107 differentially expressed MYC-signature genes in PPKS/M relative to PPKS/NT tumors. C, Dotplot of normalized enrichment score (NES) and FDR of MYC-associated gene sets in PPKS/M relative to PPKS/NT tumors. D, GSEA categories enriched in PPKS/M relative to PPKS/NT tumors. E, Dotplot of immune response–associated gene sets negatively enriched in PPKS/M relative to PPKS/NT tumors. F, IHC analysis of CD45+ immune cells, F4/80+ macrophages, CD8+ T cells, CD4+ T cells, and CD19+ B cells in PPKS/M and control (PPKY/LV and PPKS/NT) tumors. Scale bar, 100 μm. G, qIHC of CD45+ immune cells. H, qIHC of F4/80+ macrophages. I, qIHC of CD8+ T cells. J, qIHC analysis of CD4+ T. K, qIHC of CD19+ B cells. L, Heatmap and unsupervised hierarchical clustering of tumors. M, CMScaller subtype by tumor cohort. G–K, Significance tested with two-tailed the Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Figure 5.

Transcriptomic reprogramming in Myc-activated lung tumors from the CRISPRa/SAM model. A, Heatmap of differentially expressed MYC network members in PPKS/M relative to PPKS/NT tumors, Myc/Mycn/Mycl (red). B, Heatmap of 107 differentially expressed MYC-signature genes in PPKS/M relative to PPKS/NT tumors. C, Dotplot of normalized enrichment score (NES) and FDR of MYC-associated gene sets in PPKS/M relative to PPKS/NT tumors. D, GSEA categories enriched in PPKS/M relative to PPKS/NT tumors. E, Dotplot of immune response–associated gene sets negatively enriched in PPKS/M relative to PPKS/NT tumors. F, IHC analysis of CD45+ immune cells, F4/80+ macrophages, CD8+ T cells, CD4+ T cells, and CD19+ B cells in PPKS/M and control (PPKY/LV and PPKS/NT) tumors. Scale bar, 100 μm. G, qIHC of CD45+ immune cells. H, qIHC of F4/80+ macrophages. I, qIHC of CD8+ T cells. J, qIHC analysis of CD4+ T. K, qIHC of CD19+ B cells. L, Heatmap and unsupervised hierarchical clustering of tumors. M, CMScaller subtype by tumor cohort. G–K, Significance tested with two-tailed the Student t test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Close modal

To infer the effect of Myc-activation on downstream targets, we probed the expression of a list of 134 MYC-signature genes with evidence of direct Myc-binding (see Supplementary Table S4; refs. 30–32). We found that 100 of these genes (75%) were significantly overexpressed in PPKS/M relative to PPKS/NT tumors, whereas 7 genes (5%) were significantly downregulated (Fig. 5B). Using the expression values for these genes, we calculated an MYC signature score (see Materials and Methods) for each tumor that was significantly higher in PPKS/M tumors compared with all other groups (Supplementary Fig. S5A).

To determine whether we could detect an Myc-driven tumor phenotype, we performed GSEA on genes differentially expressed in PPKS/M relative to PPKS/NT tumors. GSEA revealed strong enrichment for multiple sets of MYC target genes (Fig. 5C; Supplementary Fig. S5B). In addition, GSEA showed strong positive enrichment for canonical MYC-associated biological processes in PPKS/M tumors, including translation/protein biosynthesis, replication/cell cycle, DNA damage response/repair, and oxidative phosphorylation/mitochondrial function (Fig. 5D; Supplementary Fig. S5C–S5F and Supplementary Table S5). Importantly, we found that a large number of immune response-related gene sets were negatively enriched (Fig. 5E). Consistent with our GSEA results, we found that PPKS/M tumor cell lines displayed significantly increased baseline DNA damage compared with controls, as indicated by an increase in nuclear γH2AX foci (Supplementary Fig. S5G and S5H). We also found that PPKS/M cells were significantly more sensitive to oxaliplatin relative to PPKS/NT cells, as indicated by lower IC50 values, 3.4 and 5.4 μmol/L, respectively (Supplementary Fig. S5I).

In line with our GSEA results, there is accumulating evidence that Myc drives an immunosuppressive microenvironment in non–small cell lung cancer and other tumor entities (18–21, 33). To investigate this potential link, we set out to decipher the immune composition in our PPKS/M lung tumors using quantitative IHC (qIHC). Next to a significant overall decrease in immune cells (CD45+) in Myc-activated tumors, we found that the population of cytotoxic T cells (CD8+), T helper-cells (CD4+), B cells (CD19+), and macrophages (F4/80+) was significantly decreased compared with controls (Fig. 5FK). We also found that Myc-activated tumors expressed lower levels of several Class I MHC genes (Supplementary Fig. S5J). These observations reveal a profound effect of Myc activation in driving a less immunogenic, or “immune cold,” and likely immunosuppressive tumor microenvironment in our model. Taken together, these results show that our model faithfully recapitulates numerous MYC-associated features and indicates the wide-reaching potential of our approach.

Transcriptomic subtyping of murine and human lung adenocarcinomas reveals MYC as a driver of the LuAd2 subtype

Recently, Jang and colleagues (22) classified a cohort of human lung adenocarcinoma tumors into three subtypes (LuAd1–3) with distinct transcriptomic and immune-regulatory features, and potential immunotherapy-response predictive power. LuAd1 was found to represent an “immune-hot” tumor subtype with potentially favorable response to immune therapy, whereas LuAd2 and LuAd3 represented distinct “immune-cold” tumor subtypes. However, the transcriptomic drivers of these subtypes remain to be elucidated. To characterize our mouse tumors using an analogous approach, we subtyped the tumors with CMScaller (28) using lists of subtype-specific, overexpressed gene orthologs from the Jang and colleagues (22) study (Supplementary Table S6). Tumor subtyping revealed starkly different transcriptomic profiles between individual mouse tumors (Fig. 5L). Although all Yap1-activated tumors were subtyped as LuAd1, strikingly, all Myc-activated tumors were subtyped as LuAd2 and clustered separately from the other tumors (Fig. 5L and M). On the basis of this observation, we hypothesized that Myc is a potential driver of the LuAd2 subtype.

To expand this analysis and to investigate a potential link between MYC signaling and tumor subtype in human lung adenocarcinomas, we applied a similar approach to stratify the 585 lung adenocarcinoma cases in the TCGA-LUAD cohort. In summary, 206 (35%), 256 (44%), and 93 (16%) patient tumors were subtyped as LuAd1, LuAd2, and LuAd3, respectively, whereas 30 (5%) patient tumors did not classify as any subtype (Fig. 6A; Supplementary Fig. S6A, Supplementary Table S6).

Figure 6.

Transcriptomic subtyping of lung adenocarcinoma tumors reveals enrichment of MYC signature in LuAd2 patient tumors. A, Heatmap of subtype-specific gene expression in TCGA-LUAD patient tumors by CMScaller subtype (LuAd1/LuAd2/LuAd3) and selected immune regulatory genes, MHC molecules, MYC signaling network members, maximum MYC/N/L, MYC signature, IR signature, and MYC, MYCN, and MYCL copy-number variation. B, GSEA of MYC-associated gene sets in LuAd2 relative to all other TCGA-LUAD tumors. C, Dot plot showing normalized enrichment score (NES) and FDR of MYC-associated gene sets in LuAd2 tumors. D, Normalized MYC expression in LuAd1, LuAd2, and LuAd3 tumors. E, LUAD subtype distribution as function of MYC copy-number status, χ2 test. F, Maximum normalized expression of MYC, MYCN, or MYCL from each tumor by subtype. G, Kaplan–Meier survival plot of the TCGA-LUAD dataset by MYC-signature; LOW (<25 percentile, n = 146), MED (25–75 percentile, n = 292), HIGH (>75 percentile, n = 147), pairwise Mantel–Cox (log rank) test. H, MYC signature score by subtype. D, F, and H, Significance tested using one-way ANOVA with the Tukey multiple comparison test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Figure 6.

Transcriptomic subtyping of lung adenocarcinoma tumors reveals enrichment of MYC signature in LuAd2 patient tumors. A, Heatmap of subtype-specific gene expression in TCGA-LUAD patient tumors by CMScaller subtype (LuAd1/LuAd2/LuAd3) and selected immune regulatory genes, MHC molecules, MYC signaling network members, maximum MYC/N/L, MYC signature, IR signature, and MYC, MYCN, and MYCL copy-number variation. B, GSEA of MYC-associated gene sets in LuAd2 relative to all other TCGA-LUAD tumors. C, Dot plot showing normalized enrichment score (NES) and FDR of MYC-associated gene sets in LuAd2 tumors. D, Normalized MYC expression in LuAd1, LuAd2, and LuAd3 tumors. E, LUAD subtype distribution as function of MYC copy-number status, χ2 test. F, Maximum normalized expression of MYC, MYCN, or MYCL from each tumor by subtype. G, Kaplan–Meier survival plot of the TCGA-LUAD dataset by MYC-signature; LOW (<25 percentile, n = 146), MED (25–75 percentile, n = 292), HIGH (>75 percentile, n = 147), pairwise Mantel–Cox (log rank) test. H, MYC signature score by subtype. D, F, and H, Significance tested using one-way ANOVA with the Tukey multiple comparison test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Close modal

Interestingly, we found that multiple MYC-associated gene sets were strongly enriched in LuAd2 tumors over all other cases (Fig. 6B and C). We found that LuAd2 tumors expressed significantly more MYC relative to LuAd3, and MYC copy-number amplified tumors were significantly enriched for the LuAd2 subtype (Fig. 6A and DE). When we considered the maximum expression value for MYC, MYCN and MYCL for each tumor, we found that LuAd2 expressed significantly higher levels compared both other groups (Fig. 6A and F), consistent with a direct link between the MYC/MYCN/MYCL signaling axis and LuAd2.

Next, we defined an MYC signature score based on the expression of 134 well-established MYC-target genes (30–32). Stratification of the TCGA-cohort into 3 cohorts revealed that a high MYC signature score was associated with significantly shorter survival relative to MYC signature medium and low tumors (median survival 40 (high), 50 (medium) and 59 months (low), Fig. 6G). Consistent with our hypothesis, we found that LuAd2 tumors displayed a strongly elevated MYC signature score (Fig. 6A and H). Taken together, transcriptomic subtyping of our PPKS/M tumors and human lung adenocarcinoma cohorts (TCGA) allowed us to identify MYC as a potential driver for the LuAd2 subtype, warranting further investigation of the link between MYC and tumor immune landscape.

MYC signaling as a driver of an immunosuppressive lung adenocarcinoma subtype may predict immune therapy response

Consistent with Jang and colleagues (22), we found that LuAd2 and LuAd3 tumors in the TCGA-LUAD dataset express significantly lower levels of immune regulatory factors, MHC molecules (Figs. 6A, 7A and B), and immune response score (IRscore) relative to LuAd1 tumors (Fig. 7C; Supplementary Table S6). Moreover, the LuAd2 subtype was associated with significantly shorter survival as compared with LuAd1 (median survival 38.47 and 71.42 months, P < 0.0001, Fig. 7D), whereas no significant difference in survival was observed between LuAd2 and LuAd3 nor between LuAd3 and LuAd1. We were thus able to validate the major findings from the Jang and colleagues (22) study across a much larger cohort of patient tumors.

Figure 7.

MYC signature and immune response. A, Normalized expression of immune regulatory factors by subtype in the TCGA-LUAD dataset. B, MHC factor expression by subtype. C, IR signature by subtype. D, Kaplan–Meier survival plot of the TCGA-LUAD dataset, by subtype. Pairwise, the Mantel–Cox (log rank) test. E, Correlation plot of IR signature and MYC signature in the TCGA-LUAD dataset. Pearson correlation coefficient r = −0.42, P < 0.0001. A–C, Significance tested using one-way ANOVA with the Tukey multiple comparison test. ****, P ≤ 0.0001.

Figure 7.

MYC signature and immune response. A, Normalized expression of immune regulatory factors by subtype in the TCGA-LUAD dataset. B, MHC factor expression by subtype. C, IR signature by subtype. D, Kaplan–Meier survival plot of the TCGA-LUAD dataset, by subtype. Pairwise, the Mantel–Cox (log rank) test. E, Correlation plot of IR signature and MYC signature in the TCGA-LUAD dataset. Pearson correlation coefficient r = −0.42, P < 0.0001. A–C, Significance tested using one-way ANOVA with the Tukey multiple comparison test. ****, P ≤ 0.0001.

Close modal

We found that the IRscore was significantly negatively correlated with the MYC signature score across the TCGA-LUAD cohort (Pearson r = −0.42, P < 0.0001), consistent with an antagonistic relationship between MYC signaling and factors important for response to immune therapy (Fig. 7E). We also implemented our analysis on an independent dataset from OncoSG (169 patients). Within this dataset, we were able to recapitulate our main findings from the TCGA dataset; LuAd2 tumors displayed significantly higher MYC signature scores and lower IRscores as compared with LuAd1, and the MYC signature score again predicted survival (Supplementary Fig. S6B–S6I)

Overall, these results, combined with the results from our mouse model, strongly implicate MYC signaling as a driver of the “immune cold” LuAd2 subtype. This highlights the potential of MYC-targeted pharmacological intervention to improve patient responses to immune checkpoint blockade in this tumor subtype.

Dysregulation of gene expression is a key characteristic of many human diseases, including cancer. Introducing controlled gene perturbations in animal models through genetic engineering is necessary to faithfully recapitulate human disease. The groundbreaking developments in genetic engineering of the last decade have opened new avenues for more flexible and sophisticated models. Here, we have demonstrated that our CRISPRa/LSL-SAM mouse model is a powerful and versatile tool for programmable gene activation and oncogene evaluation in vitro and in vivo. We have shown gene activation in primary cells ex vivo, and transplantation into syngeneic mice to generate tumors. This approach could straightforwardly be expanded to ex vivo or in vivo screening using pools of gene-targeting guides. Importantly, we detected only moderate levels of transcriptional activation, that is, only marginally supraphysiological levels. This is in stark contrast with conventional, viral promoter driven/transgene approaches where overexpression can reach 100-fold, potentially resulting in artificial rather than physiologically relevant effects.

We have also demonstrated that gene activation can be triggered directly in vivo through lentiviral transduction and that oncogene activation can drive dramatic differences in tumor phenotype, recapitulating observations from human patients. We developed reliable protocols for in-house high titer lentivirus production, and titer and transduction efficiency quantification. The direct in vivo transduction approach can likely also be adapted to allow for limited-scale oncogene screening through the use of pooled lentiviral guides, efforts that we are currently pursuing.

By activating the Myc oncogene concurrently with oncogenic Kras activation and P53 loss, we recapitulated important MYC-associated features of human lung tumors. The results from our model also generated the hypothesis of a direct link between MYC signaling, an immunosuppressive tumor microenvironment and a clinically relevant tumor subtype in lung adenocarcinoma (LuAd2). Through cross-species validation, our present work highlighted a significant cohort of human lung adenocarcinoma patient tumors classified as LuAd2 and characterized by a pronounced MYC signature and immune suppression. This MYC signature high/LuAd2 subtype patient cohort may display enhanced response to combined MYC/immune-targeting therapies.

F.I. Thege reports grants from Ben and Rose Cole Charitable Pria Foundation and Cancer Prevention and Research Institute of Texas (CPRIT) during the conduct of the study. A. Maitra reports other support from Cosmos Wisdom Biotechnology and Thrive Earlier Detection (Exact Sciences company) as well as personal fees from Freenome and Tezcat Biotechnology outside the submitted work. S.M. Wörmann reports grants from Ben and Rose Cole Charitable Pria Foundation, German Cancer Aid, the Mildred-Scheel-Postdoctoral Fellow Program, MD Anderson GI SPORE CEP, CPRIT, and APA Foundation during the conduct of the study. No disclosures were reported by the other authors.

F.I. Thege: Conceptualization, software, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. D.N. Rupani: Investigation, writing–review and editing. B. Brahmendra Barathi: Investigation. S.L. Manning: Investigation. A. Maitra: Resources, supervision, funding acquisition, writing–review and editing. A.D. Rhim: Resources, funding acquisition. S.M. Wörmann: Conceptualization, investigation, methodology, writing–original draft, writing–review and editing.

F.I. Thege and S.M. Wörmann are supported by the Ben and Rose Cole Charitable Pria Foundation. S.M. Wörmann is supported by German Cancer Aid, the Mildred–Scheel–Postdoctoral Fellow Program, the MD Anderson GI SPORE CEP. A.D. Rhim and A. Maitra are supported by the MD Anderson Moonshot Program in Pancreatic Cancer. A.D. Rhim is supported by the V Foundation, Doris Duke Foundation, Andrew Sabin Family Foundation, Cockrell Family Foundation, NCI (MD Anderson SPORE), and CPRIT. A. Maitra is supported by the Khalifa bin Zayed Foundation. The MDACC Flow cytometry core facility is supported by the NCI Cancer Center support grant P30CA16672. The MDACC Genetically Engineered Mouse Facility (GEMF) is supported by NCI Cancer Center Support grant P30CA016672 and NCI R50CA211121. The authors thank members of the A.D. Rhim and A. Maitra laboratories for critical reading of the article.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Jinek
M
,
Chylinski
K
,
Fonfara
I
,
Hauer
M
,
Doudna
JA
,
Charpentier
E
.
A programmable dual-RNA–guided DNA endonuclease in adaptive bacterial immunity
.
Science
2012
;
337
:
816
21
.
2.
Cong
L
,
Ran
FA
,
Cox
D
,
Lin
S
,
Barretto
R
,
Habib
N
, et al
.
Multiplex genome engineering using CRISPR/Cas systems
.
Science
2013
;
339
:
819
23
.
3.
Mali
P
,
Yang
L
,
Esvelt
KM
,
Aach
J
,
Guell
M
,
DiCarlo
JE
, et al
.
RNA-guided human genome engineering via Cas9
.
Science
2013
;
339
:
823
6
.
4.
Platt
RJ
,
Chen
S
,
Zhou
Y
,
Yim
MJ
,
Swiech
L
,
Kempton
HR
, et al
.
CRISPR–Cas9 knockin mice for genome editing and cancer modeling
.
Cell
2014
;
159
:
440
55
.
5.
Chiou
SH
,
Winters
IP
,
Wang
J
,
Naranjo
S
,
Dudgeon
C
,
Tamburini
FB
, et al
.
Pancreatic cancer modeling using retrograde viral vector delivery and in vivo CRISPR/Cas9-mediated somatic genome editing
.
Genes Dev
2015
;
29
:
1576
85
.
6.
Maresch
R
,
Mueller
S
,
Veltkamp
C
,
Ollinger
R
,
Friedrich
M
,
Heid
I
, et al
.
Multiplexed pancreatic genome engineering and cancer induction by transfection-based CRISPR/Cas9 delivery in mice
.
Nat Commun
2016
;
7
:
10770
.
7.
Mueller
S
,
Engleitner
T
,
Maresch
R
,
Zukowska
M
,
Lange
S
,
Kaltenbacher
T
, et al
.
Evolutionary routes and KRAS dosage define pancreatic cancer phenotypes
.
Nature
2018
;
554
:
62
8
.
8.
Konermann
S
,
Brigham
MD
,
Trevino
AE
,
Joung
J
,
Abudayyeh
OO
,
Barcena
C
, et al
.
Genome-scale transcriptional activation by an engineered CRISPR–Cas9 complex
.
Nature
2015
;
517
:
583
8
.
9.
Chavez
A
,
Tuttle
M
,
Pruitt
BW
,
Ewen-Campen
B
,
Chari
R
,
Ter-Ovanesyan
D
, et al
.
Comparison of Cas9 activators in multiple species
.
Nat Methods
2016
;
13
:
563
7
.
10.
Joung
J
,
Konermann
S
,
Gootenberg
JS
,
Abudayyeh
OO
,
Platt
RJ
,
Brigham
MD
, et al
.
Genome-scale CRISPR-Cas9 knockout and transcriptional activation screening
.
Nat Protoc
2017
;
12
:
828
63
.
11.
Fidanza
A
,
Lopez-Yrigoyen
M
,
Romano
N
,
Jones
R
,
Taylor
AH
,
Forrester
LM
.
An all-in-one UniSam vector system for efficient gene activation
.
Sci Rep
2017
;
7
:
6394
.
12.
Hunt
C
,
Hartford
SA
,
White
D
,
Pefanis
E
,
Hanna
T
,
Herman
C
, et al
.
Tissue-specific activation of gene expression by the Synergistic Activation Mediator (SAM) CRISPRa system in mice
.
Nat Commun
2021
;
12
:
2770
.
13.
Naldini
L
,
Blomer
U
,
Gallay
P
,
Ory
D
,
Mulligan
R
,
Gage
FH
, et al
.
In vivo gene delivery and stable transduction of nondividing cells by a lentiviral vector
.
Science
1996
;
272
:
263
7
.
14.
Kumar
MS
,
Erkeland
SJ
,
Pester
RE
,
Chen
CY
,
Ebert
MS
,
Sharp
PA
, et al
.
Suppression of non–small cell lung tumor development by the let-7 microRNA family
.
Proc Natl Acad Sci U S A
2008
;
105
:
3903
8
.
15.
Jackson
EL
,
Willis
N
,
Mercer
K
,
Bronson
RT
,
Crowley
D
,
Montoya
R
, et al
.
Analysis of lung tumor initiation and progression using conditional expression of oncogenic K-ras
.
Genes Dev
2001
;
15
:
3243
8
.
16.
Meuwissen
R
,
Linn
SC
,
van der Valk
M
,
Mooi
WJ
,
Berns
A
.
Mouse model for lung tumorigenesis through Cre/lox controlled sporadic activation of the K-Ras oncogene
.
Oncogene
2001
;
20
:
6551
8
.
17.
DuPage
M
,
Dooley
AL
,
Jacks
T
.
Conditional mouse lung cancer models using adenoviral or lentiviral delivery of Cre recombinase
.
Nat Protoc
2009
;
4
:
1064
72
.
18.
Kortlever
RM
,
Sodir
NM
,
Wilson
CH
,
Burkhart
DL
,
Pellegrinet
L
,
Brown Swigart
L
, et al
.
Myc cooperates with ras by programming inflammation and immune suppression
.
Cell
2017
;
171
:
1301
15
.
19.
Topper
MJ
,
Vaz
M
,
Chiappinelli
KB
,
DeStefano Shields
CE
,
Niknafs
N
,
Yen
RC
, et al
.
Epigenetic therapy ties MYC depletion to reversing immune evasion and treating lung cancer
.
Cell
2017
;
171
:
1284
300
.
20.
Casey
SC
,
Tong
L
,
Li
Y
,
Do
R
,
Walz
S
,
Fitzgerald
KN
, et al
.
MYC regulates the antitumor immune response through CD47 and PD-L1
.
Science
2016
;
352
:
227
31
.
21.
Han
H
,
Jain
AD
,
Truica
MI
,
Izquierdo-Ferrer
J
,
Anker
JF
,
Lysy
B
, et al
.
Small-molecule MYC inhibitors suppress tumor growth and enhance immunotherapy
.
Cancer Cell
2019
;
36
:
483
97
.
22.
Jang
HJ
,
Lee
HS
,
Ramos
D
,
Park
IK
,
Kang
CH
,
Burt
BM
, et al
.
Transcriptome-based molecular subtyping of non–small cell lung cancer may predict response to immune checkpoint inhibitors
.
J Thorac Cardiovasc Surg
2020
;
159
:
1598
610
.
23.
Rhim
AD
,
Mirek
ET
,
Aiello
NM
,
Maitra
A
,
Bailey
JM
,
McAllister
F
, et al
.
EMT and dissemination precede pancreatic tumor formation
.
Cell
2012
;
148
:
349
61
.
24.
Ai
J
,
Wormann
SM
,
Gorgulu
K
,
Vallespinos
M
,
Zagorac
S
,
Alcala
S
, et al
.
Bcl3 couples cancer stem cell enrichment with pancreatic cancer molecular subtypes
.
Gastroenterology
2021
;
161
:
318
32
.
25.
Boj
SF
,
Hwang
CI
,
Baker
LA
,
Chio
II
,
Engle
DD
,
Corbo
V
, et al
.
Organoid models of human and mouse ductal pancreatic cancer
.
Cell
2015
;
160
:
324
38
.
26.
Thege
FI
,
Gruber
CN
,
Cardle
II
,
Cong
SH
,
Lannin
TB
,
Kirby
BJ
.
anti-EGFR capture mitigates EMT- and chemoresistance-associated heterogeneity in a resistance-profiling CTC platform
.
Anal Biochem
2019
;
577
:
26
33
.
27.
Wörmann
SM
,
Zhang
A
,
Thege
FI
,
Cowan
RW
,
Rupani
DN
,
Wang
R
, et al
.
APOBEC3A drives deaminase domain-independent chromosomal instability to promote pancreatic cancer metastasis
.
Nature Cancer
2021
;
2
:
1338
56
.
28.
Eide
PW
,
Bruun
J
,
Lothe
RA
,
Sveen
A
.
CMScaller: an R package for consensus molecular subtyping of colorectal cancer pre-clinical models
.
Sci Rep
2017
;
7
:
16618
.
29.
Schaub
FX
,
Dhankani
V
,
Berger
AC
,
Trivedi
M
,
Richardson
AB
,
Shaw
R
, et al
.
Pan-cancer alterations of the MYC oncogene and its proximal network across the cancer genome atlas
.
Cell Syst
2018
;
6
:
282
300
.
30.
Kidder
BL
,
Yang
J
,
Palmer
S
.
Stat3 and c-Myc genome-wide promoter occupancy in embryonic stem cells
.
PLoS ONE
2008
;
3
:
e3932
.
31.
Kim
J
,
Chu
J
,
Shen
X
,
Wang
J
,
Orkin
SH
.
An extended transcriptional network for pluripotency of embryonic stem cells
.
Cell
2008
;
132
:
1049
61
.
32.
Chandriani
S
,
Frengen
E
,
Cowling
VH
,
Pendergrass
SA
,
Perou
CM
,
Whitfield
ML
, et al
.
A core MYC gene expression signature is prominent in basal-like breast cancer but only partially overlaps the core serum response
.
PLoS ONE
2009
;
4
:
e6693
.
33.
Masso-Valles
D
,
Beaulieu
ME
,
Soucek
L
.
MYC, MYCL, and MYCN as therapeutic targets in lung cancer
.
Expert Opin Ther Targets
2020
;
24
:
101
14
.

Supplementary data