Fig 2: Effects of GSK5182 on iodide handing genes of BCPAP cells. (A) Expression level of TG, TSHR, TPO, and Pax-8 protein in GSK5182-treated cells. (B–E) Graph showing results of quantitative analysis of iodide metabolism-related genes by Image J. BCPAP cells were treated with GSK518 for 24 h, followed by collection of RNA and protein samples. *, p < 0.05. NS, not significant. Data are the mean ± SD of three samples per group.

Fig 3: Systematic molecular definition of lineage factor inhibition resistance across cancer lineages(A) Schematic outlining the analytical pipeline used to identify lineage factor inhibition-resistant cell lines across cancer types, and uncover shared molecular features.(B) Frequency distribution of the maximum possible LD (i.e., most negative) score for each TF. For example, PAX8 has the strongest lineage-specific dependency score in RCC and so the PAX8 LD score in the RCC context is plotted here.(C) Frequency distribution of the CERES score for filtered LDs, in each cell line of their respective lineage. Distribution statistically deviated from normal determined using Shapiro Test.(D) Tukey plot of the average CERES score of each LD in their respective lineage. Sensitive and resistant lines are identified using the cutoff identified in (B) and (C), average CERES > -0.45. For boxplots, center line shows the median, the box bounds represent the first and third quartiles, and the whiskers extend to the highest and lowest values no higher of lower than 1.5 * IQR. ALL: acute lymphoblastic leukemia, AML: acute myeloid leukemia, EWS: Ewing sarcoma, RCC: renal cell carcinoma, NHL: non-Hodgkin lymphoma, NB: neuroblastoma, MM: multiple myeloma, Mel: melanoma, RMS: rhabdomyosarcoma, EAC: endometrial adenocarcinoma.(E) Summary of acquired dependency analysis across different cancer lineages. FDR calculated using permutation-based statistics. Highlighted points satisfy FDR<0.1.(F) Boxplot of the CERES score of lineage-sensitive (i.e., lineage factor dependent, n = 30) and resistant (i.e., lineage factor independent, n = 19) cell lines from the melanoma lineage for TP53 KO. FDR calculated using permutation-based statistics. For boxplots, center line shows the median, the box bounds represent the first and third quartiles, and the whiskers extend to the highest and lowest values.(G) Stacked bar plot representing the proportion of TP53 mutations in lineage-sensitive (n = 30) and resistant (n = 19) cell lines from the melanoma lineage. p-value generated using Fisher’s exact test.(H) Boxplot of the CERES score of lineage-sensitive (i.e., lineage factor dependent, n = 6) and resistant cell (i.e., lineage factor independent, n = 4) lines from the rhabdomyosarcoma lineage for MYC KO. FDR calculated using permutation-based statistics. For boxplots, center line shows the median, the box bounds represent the first and third quartiles, and the whiskers extend to the highest and lowest values. See also Figure S7.

Fig 4: SMARCB1 loss facilitates resistance to lineage factor inhibition in ccRCC cells(A) A schematic of a pooled CRISPR-Cas9-based loss-of-function screen using an sgRNA library targeting chromatin regulators to identify genes capable of modulating the dependency of ccRCC cell lines on PAX8 inhibition.(B and C) Changes in sgRNA abundance over time, measured by calculating beta scores using the top three enriched sgRNAs per gene relative to day 0, from two technical replicates. (B) The beta scores for the control arm of the screen versus the pooled experimental arm (P81/2). (C) The beta scores for each of the experimental arms of the screen. Highlighted points have a beta score <-0.5 or >0.5 and a p value <0.05. p-value was calculated by permutation based approach using MAGeCK.(D) Genetic dependency data for 946 cell lines from the DepMap project. Distribution of CERES scores for the gene SMARCB1 and an example non-essential gene (CNBD1). Kruskal-Wallis test.(E and F) Competitive proliferation assays using 786-M1A Cas9 clone 6 (786-M1A-C6) cells expressing combinations of sgRNAs (Ctrl or S11/2) and inducible shRNAs (Ctrl or P81/2), competed against cells with a PAX8 KD (P81/2). Three technical replicates per condition. (E) Doxycycline was added at day 0 (acute, A), (F) cells were pre-treated for ~1 month before the start of the assay (midterm, MT). Error bars are SD. Kruskal-Wallis test.(G) Escaper assay with three technical replicates using 786-M1A-C6 cells, showing the percentage of cells expressing the fluorophore dsRed as a measure of PAX8 shRNA expression for cells pre-cultured on doxycycline for ~1 month (MT), normalized to day 0. Cells were sorted at the beginning of the assay to ensure a starting point of 100% dsRed. Error bars are SD.(H) Hematoxylin and eosin (H&E) staining of tumors formed by Ctrl.Ctrl and P81S11(LT) cells, respectively.(I) Boxplot of PAX8-centered CERES dependency score for VHL-mutant RCC lines from the DepMap project, which are either WT (n = 13) or mutant for ARID1A (n = 1) or SMARCB1 (n = 3). Kruskal-Wallis test. For boxplots, center line shows the median, the box bounds represent the first and third quartiles, and the whiskers extend to the highest and lowest values. See also Figures S1 and S2.

Fig 5: Acquired requirement of rare transcriptional dependencies in lineage factor inhibition-resistant ccRCC cells(A) Schematic overview of a CRISPR-Cas9 screen to look for new transcriptional dependencies in PAX8 KD/SMARCB1 KO cells.(B) The beta scores for the control arm (Ctrl.Ctrl) of the screen versus the pooled experimental arm (P81S11/2(LT)). Beta scores were calculated using the fold change of the top three depleted sgRNAs per gene relative to the plasmid library, from two replicates. p-value calculated by permutation method using MAGeCK.(C) Genetic dependency data from the DepMap project. IRF2 centered CERES dependency score of multiple myeloma (MM) cell lines (n = 21) versus cells from all other lineages (n = 767). Kruskal-Wallis test.(D) GSEA plot of interferon-gamma response signature from mSigDB hallmarks collection for P81Ctrl(A) vs. P81S11/2(LT).(E) Genetic dependency data from the DepMap project. BHLHE40-centered CERES dependency score for RCC cell lines (n = 23) versus cells from all other lineages (n = 765). Kruskal-Wallis test.(F) VHL-mutant ccRCC DepMap cell lines ranked by BHLHE40-centered CERES dependency score. SMARCB1 and ARID1A mutant cell lines from Figure 1I.(G) Genetic dependency data from the DepMap project. ZNFX1-centered CERES dependency scores across 25 lineages, with =10 cell lines per lineage. CCA: cholangiocarcinoma, ALL: acute lymphoblastic leukemia, AML: acute myeloid leukemia, EWS: Ewing sarcoma, BRC: breast carcinoma, BRDC: breast ductal carcinoma, Gli: Glioma, CRC: colorectal adenocarcinoma, ESCC: esophageal squamous cell carcinoma, GA: gastric adenocarcinoma, RCC: renal cell carcinoma, HCC: hepatocellular carcinoma, LMT: lung mesothelioma, NSCLC: non-small-cell lung cancer, SCLC: small-cell lung cancer, NHL: non-Hodgkin lymphoma, OA: ovarian adenocarcinoma, EPC: exocrine pancreatic cancer, NB: neuroblastoma, MM: multiple myeloma, Mel: melanoma, RMS: rhabdomyosarcoma, UATN: upper aerodigestive tract neoplasm, BC: bladder carcinoma, EAC: endometrial adenocarcinoma.(H) MYC expression as determined by RNA-seq. Adjusted P-value by DESeq2.(I) MYC expression as determined by quantitative RT-PCR in the indicated cell lines after IRF2 and BHLHE40 inactivation by CRISPR-Cas9. M1A clone 11 is a Cas9-expressing clonal derivative of 786-M1A cells. p-values calculated using one-way ANOVA. See also Figure S6.