(A) Design of shRNA screen. T1 and T2 show 3 and 6 weeks in vitro proliferation, respectively. T3 shows 3 weeks of subcutaneous growth. Colored bars represent the relative ratio of the MFI of a particular barcode corresponding to a shRNA at T1/T2/T3 over T0. Light blue bar depicts a theoretical barcode negatively selected over time. (B) Quantitation of each shRNA at T2 compared with T0. MFI measurements are average of 3 independent experiments done for each of the 2 cell lines at each time point. x axis shows log₂ of average mean MFI for each shRNA. y axis shows log₂ of fold change of MFI between the 2 time points. Red squares indicate shRNAs negatively selected as described in text, with at least 1 other negatively selected shRNA against the same gene. (C) Knockdown of intended target for each shRNA identified as negatively selected using rtPCR. Results show relative mRNA expression for each gene compared with a control shRNA. Results for knockdown of Kras are also shown. (D) Proliferation of LKR13 cells infected with shRNAs for target genes that showed on-target effect as demonstrated in C. Cells were place in 96-well plates and analyzed using the MTT assay at day 7 after selection. Results are the average of 3 independent measurements. Error bars indicate mean ± SD.

(A) Primers for rtPCR were designed to amplify exons 8 (F) and 9 (R) of wild-type and conditional Wt1 alleles, and span intronic sequence. (B) rtPCR results demonstrating deletion of full-length Wt1 in Wt1^Δ/Δ MEFs. (C) Cell-cycle analysis of MEFs 7 days after infection with AdGFP (control) or AdCre. Numbers represent the percentage of cells in G₁ (lower left), S (upper left) and G₂/M (lower right) phase. (D) 3T3 assay of Kras^LSL-G12D/+, Kras^G12D/+,Wt1^Δ/Δ, and Kras^G12D/+;Wt1^Δ/Δ MEFs. Wt1^f/f and Kras^LSL-G12D/+;Wt1^f/f MEFs were also analyzed and performed similarly to Kras^LSL-G12D/+ and Wt1^Δ/Δ MEFs (data not shown). (E) Micrographs and quantitation of SA-βgal staining of MEFs of indicated genotypes. Staining was performed 48 hours after plating of cells and 11 days after AdCre or AdGFP infection. Scale bars: 200 μm. (F) Immunofluorescence demonstrating increased levels of Pml-containing PNBs in Kras^G12D/+;Wt1^Δ/Δ MEFs. Scale bars: 100 μm (large images); 20 μm (insets). (G) Quantification of BrdU-positive cells 10 days after infection with AdCre. (H) Quantification of histone H3–positive cells 10 days after infection with AdCre. P values are for a 2-tailed t test. Error bars for all panels show mean ± SD.

(A) Western blot showing WT1 suppression by 2 independent shRNAs in nuclear lysates of wild-type RAS NSCLC cells, NCI-H1568 (H1568), and mutant RAS NSCLC cells, NCI-H23 (H23). (B) Cell viability analysis of human NSCLC cell lines infected with WT1 shRNAs as measured by an MTT assay at day 5 after selection. RAS wild-type cell lines: NCI-H522, NCI-H1437, NCI-H1568, NCI-H1650, NCI-H1975, and NCI-H2126. RAS mutant cell lines: NCI-H23, NCI-H358, NCI-H441, NCI-H460, NCI-H727, NCI-H1299, NCI-H2009, and NCI-A549. The same WT1 shRNAs were used for all experiments (squares: WT1 shRNA1; circles: WT1 shRNA2). Data points indicate percentage of cell viability (percentage of the ratio of WT1 shRNA–infected cells over GFP shRNA–infected cells). (C) Cell viability of pancreatic cancer cell lines with either wild-type (HPAF-II) or mutant KRAS (ASPC-1) after WT1 knockdown with the same shRNAs used in B. Results were normalized against cell viability of control cells transduced with a GFP shRNA. Error bars show mean ± SD. (D) Annexin V analysis of NSCLC cell lines infected with 2 WT1 shRNAs or a GFP shRNA. Wild-type (NCI-H1568, NCI-H1975, and NCI-H2126) and mutant (NCI-H23, NCI-A549, and NCI-H1299) RAS NSCLC cell lines were used. Data points show difference in annexin V staining in cells infected with WT1 shRNA compared with cells infected with control GFP shRNA. (E) Percentage change of BrdU-positive cells in NSCLC cell lines analyzed in D. Data points represent percentage change of BrdU incorporating cells between cells infected with WT1 shRNAs and control cells carrying a GFP shRNA. (F) Percentage of SA-βgal–positive cells in NSCLC cells used in D. Data points indicate percentage of SA-βgal–positive cells. (G) SA-βgal staining of mutant RAS NSCLC cell lines described in D. Percentage of positive cells is shown in parentheses. Arrows point at senescence cells. Scale bars: 75 μm. All graphs are representative of at least 2 experiments. For all experiments, P values are for a 2-tailed t test.

(A) Design of MEFs validation screen. MEFs were first infected with AdCre followed by infection with lentiviral vectors containing pools of shRNAs. 5 independent MEF lines of each genotype were assessed with each shRNA pool. The colored bars represent the relative ratio of the MFI of a particular barcode corresponding to a shRNA at T1, T2, or T3 over T0. The light blue bar depicts a theoretical barcode that is negatively selected. (B) Rac1, Phb2, and Wt1 are negatively selected in Kras^LSL-G12D/+ MEFs. Box plots indicate mean ± SD of log₂ of the MFI fold change for each gene comparing T1 versus T0. P values were obtained using a 2-tailed t test. (C) Wt1 shRNAs are not negatively selected in transformed lung epithelial cell lines expressing wild-type Kras. MLE12 cells were infected with the same shRNA pools used for the MEF experiments. log₂ of the MFI fold change in the presence of 2 Wt1 shRNAs after 3 weeks of proliferation in vitro was measured and compared with the results for LKR10 and LKR13. Error bars show mean ± SD. (D) LKR13 cells were infected with shRNAs to GFP (control), Kras, and Wt1, selected for 3 days, and injected subcutaneously (n = 8) into Balb/c^nu/nu mice. Results show tumor volume at final day of experiment (day 24 after injection). Representative tumors are shown. Error bars indicate mean ± SEM. All P values are for a 2-tailed t test.

(A) Heat map of gene expression analysis of Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ MEFs compared with wild-type and Wt1^Δ/Δ MEFs. Gene sets with at least a 2-fold change and an FDR of less than 0.1 for comparison between any of the 4 genotypes are shown. Enrichment plots of (B) MSigDB gene sets of glutamine metabolism genes as well as experimentally derived gene sets downregulated by glutamate or leucine starvation and (C) MSigDB gene sets for Myc pathway genes. (D) Heat map of differentially expressed genes between Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ MEFs as determined by PAM analysis. (E) A large gene expression compendium of human lung cancer gene expression (heat map) was analyzed in order to identify patients with either high expression or low expression of Kras signature genes. After training with Kras^G12D/+;Wt1^Δ/Δ versus Kras^G12D/+ MEF data, high WT1 and low WT1 gene signature samples were identified from within the human lung carcinoma “KRAS signature high” and “KRAS signature low” groups. Kaplan-Meier analysis was used to assess survival of lung cancer patients with either high or low KRAS gene expression signatures as a function of the high/low WT1 gene signature. Log-rank P values indicate differences between the 2 groups.

(A) Representative microCT scans of tumors in Kras^LSL-G12D/+ and Kras^LSL-G12D/+;Wt1^f/f treated with AdCre and allowed to develop tumors for 16 weeks. H. heart. (B) Representative H&E-stained sections for Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ lungs (montage of ×5 magnification of lung sections). (C) Quantification of tumor area in Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ groups. (D) Dot plot showing mean number of tumors per mouse in Kras^LSLG12D/+ (n = 10) and Kras^G12D/+;Wt1^Δ/Δ (n = 13) mice. Data points indicate total number of tumors per mouse. (E) Distribution of the total number of tumors into quartiles for Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ genotypes, based on their size. Median was calculated at each quartile, and P values for the differences are indicated (P < 0.001). In addition, the percentage of the difference between the 2 groups was calculated. Percentage differences at each quartile are 17% (first), 25% (second), 33% (third), and 47% (fourth). (F) Representative images of tumors from Kras^G12D/+ and Kras^G12D/+;Wt1^Δ/Δ genotypes stained with the cell proliferation marker Ki67. Scale bars: 300 μm. Graph shows number of Ki67-positive cells per mm² of tumor. Mice were infected with AdCre for 10 weeks, and all tumors included in the analysis (Kras^G12D/+, n = 24; Kras^G12D/+;Wt1^Δ/Δ, n = 22) were confirmed to have recombination of the Wt1^f/f allele. P values are for a 2-tailed t test. Error bars indicate mean ± SEM.