Article Figures & Data
Figures
- Figure S1. Whole-genome screens in STAG2 KO lines.
(A) sgRNA sequences used to knockout STAG2 in HAP1 and RPE1 cell line backgrounds. (B) Effect of various drugs on IC50s of STAG2− lines when compared with the IC50 of STAG2+ lines. For cell lines with multiple STAG2 KO clones, IC50s of clones were averaged before being divided by the STAG2 wild-type IC50. IC50s for the H4 response to olaparib and CPT were calculated from data previously published in Bailey et al (2014). (C) Schematic of CRISPR screening protocol to assess potential STAG2 genetic interactions in three cell line backgrounds. For HAP1 screens, each of the three STAG2 knockout clones was run as one replicate and the three replicates were combined for dropout scoring. For RPE1 lines, one STAG2 KO clone (c1) was run and compared with the RPE1 WT population. (D) Average log2(fold change) gene dropout scores for previously published non-essential (top) and essential (bottom) gene sets for each screen. For H4 and RPE1 screens, both mid- and end points in the screen were sequenced. For genetic interaction analysis, RPE1 midpoint (T18) and H4 end point (T24) data were used. (E) Pearson correlation coefficients (r) of average log2(fold change) gene scores between all comparable genes in each screen and timepoint. (F) Overlap of candidate STAG2 negative genetic interactions at a cutoff of FDR < 0.5. Overlap at this cut-off is 2.30%.
- Figure 1. Whole-genome CRISPR-Cas9 knockout screen in cell lines with and without STAG2.
(A) Western blots of STAG2+ and STAG2− clones in three different backgrounds. Note that whereas HAP1 and RPE1 parents contain wild-type STAG2, H4 glioblastoma parent cells contain an endogenous STAG2 insertion that leads to protein truncation which is corrected in the knock-in (KI) cell line. (B) Effect of the PARP inhibitor talazoparib on select STAG2+ and STAG2− cells in three different backgrounds. Each line was normalized to a no drug control. Data were fitted with a non-linear IC50 curve in GraphPad Prism. (C, D, E) Comparison of STAG2+ log2-fold change scores and STAG2− log2-fold change scores in (C) HAP1, (D) RPE1, and (E) H4 backgrounds at FDR < 0.2. Candidate negative genetic interactions are in blue and candidate positive genetic interactions are in yellow. (F) Overlap of candidate STAG2 negative genetic interactions at a cut-off of qGI < −0.6 for HAP1 and residual < −1 for RPE1 and H4. % overlap at this cut-off is 2.18%.
- Figure S2. Genetic interaction between STAG1 and STAG2.
(A) Western blots of lysates from RPE1 WT and dCas9-KRAB-expressing cells after infection with non-targeting and STAG1 knockdown sgRNAs. (B) Cell cycle distributions based on propidium iodide staining of all cells (top) or pH3-positive cells (bottom) in KRAB-expressing WT and STAG2 KO lines after infection with control or STAG1 knockdown sgRNAs. Data represent averages of three independent experiments. (C) STAG1 gene effect scores from AVANA (20Q1) CRISPR screening data in cell lines with either damaging STAG2 mutations (frameshift, splice site or nonsense mutations) or other mutations (those with missense or no STAG2 mutations). Bars represent mean and standard deviation. *P < 0.05, **P < 0.005, ***P < 0.0005.
- Figure 2. STAG1 knockdown in STAG2 KO cells.
(A, B, C) Cell growth after infection of non-targeting or STAG1 knockdown sgRNAs in (A) HAP1, (B) RPE1, or (C) H4 cell lines stably expressing dCas9-KRAB. For the HAP1 lines, proliferation was determined by clonogenic assay. For H4 glioblastoma and RPE1 cells, cell growth was determined by nuclei counting. (D, E, F) Relative number of mitotic cells as analyzed by phospho-Ser10 Histone H3 (pH3) staining and flow cytometry. Data in (D, E, F) represent the average of three independent experiments. *P < 0.05, **P < 0.005, ***P < 0.0005 in a either a two-tailed, Welch’s unpaired (A, B, C) or a one-tailed, paired (D, E, F) t test.
- Figure 3. Genetic interaction between STAG2 and IREB2.
(A) Schematic of ACO1 and IRP2 function in cells. In low-iron conditions (e.g., DFO), ACO1 is in an RNA-binding form and IRP2 (the gene product of IREB2) is stabilized. Both bind iron responsive elements in target mRNAs and either stabilize them (as in the case of the iron importer TFRC) or block their translation (as with the ferritin component FTH). In high iron conditions (e.g., FAC), the IRP proteins are not needed and ACO1 is converted to an aconitase form, whereas IRP2 is degraded which destabilizes the mRNA of TFRC and allows translation of FTH. (B) Growth of HAP1 WT and STAG2 KO cells after transfection of control or IREB2 sgRNA as determined by clonogenic assay. (C) Interaction between STAG2 and IREB2 KO in H4. Cells were infected with control and IREB2 sgRNA before being normalized and re-plated for growth assessment. Cell number was determined by nuclei counting. (D) Levels of iron regulatory proteins in H4 parent and STAG2 KI cell lines including the IREB2 gene product IRP2 after infection with IREB2 and STAG1 sgRNAs. (E) Effect of sgIREB2 KO on the growth RPE1 WT and STAG2 KO cells as determined by nuclei counting. (F) Effect of ferric ammonium citrate on STAG2/IREB2 genetic interaction. Cells were infected and normalized as in (C) before 50 μM ferric ammonium citrate was added ∼24 h after re-plating. (G) Levels of iron regulatory proteins IRP2 and ACO1 in H4 parent and STAG2 KI cells after either no treatment or treatment with 100 μM DFO for ∼16 h. (H) (Top) Western blot of ACO1 levels across both STAG2+ and STAG2− cell lines in HAP1, H4 and RPE1 backgrounds. (Bottom) Amount of ACO1 as compared with HAP1 WT in three independent lysates. (I) Clonogenic growth of HAP1 WT and STAG2 KO cells after co-transfection of either vector (pcDNA3.1) or ACO1-FLAG plasmids and IREB2 sgRNAs. ns, not significant, *P < 0.05, **P < 0.005, ***P < 0.0005 in a Welch’s two-tailed t test. ns, not significant, +P < 0.005, ++P < 0.0005 in a one-way ANOVA plus TUKEY.
- Figure S3. STAG2 KO cells are sensitive to IREB2 sgRNA.
(A, B) Genetic interaction between STAG2 and IREB2 in (A) HAP1 and (B) H4 cells as seen by serial dilution. Wells were stained with crystal violet after 7–9 d. (C) Effect of IREB2 sgRNAs on its protein product IRP2 in HAP1 cells as seen by Western blot. (D) Difference in growth of H4 parent and STAG2 KI cell lines after infection with IREB2 sgRNAs as determined by clonogenic assay. (E) Protein levels of STAG2 and other cohesin components in 42 MGBA and HCT116 STAG2+/− cell lines. Similar to the H4 isogenic lines, 42 MGBA cells contain an endogenous truncating STAG2 mutation in the parent line that is corrected in the STAG2 KI cells. (F, G) Growth of (F) 42 MGBA and (G) HCT116 cell line backgrounds +/− STAG2 after infection with IREB2 sgRNAs. Cell number was determined by nuclei counting. (H) Effect of sgIREB2 on IRP2 levels in RPE1 cells. (I) IREB2 gene effect scores from DepMap (21Q2) screening data divided by damaging STAG2 mutations (nonsense, splice site, or frameshift) or other (missense or no mutation). (J) Growth of IREB2 KO cells after chronic DFO treatment. Cells were infected and plated as in Fig 3C before 0.1 μM DFO was added 24 h after re-plating. (K) Essentiality of IREB2 as related to ACO1 mRNA expression in the DepMap (20Q2) screening data. The solid line represents the smoothing spline of four knots of all data (both damaging and other mutations) generated in GraphPad Prism 8. (L) ACO1 levels in HAP1 cells after transfection with pcDNA3.1 (vector) or pcDNA3.1-ACO1-FLAG. *P < 0.05, **P < 0.005, ***P < 0.0005.
- Figure 4. STAG2 status and iron response.
(A, B, C) Cell growth of (A) HAP1, (B) H4 and (C) 42 MGBA STAG2+/− cells in the presence of the iron chelator DFO. (D) Levels of iron-related proteins including IRP targets TFRC and FTH in H4 parent and STAG2 KI cells after no treatment or treatment with 100 μM DFO or 300 μM ferric ammonium citrate.
- Figure S4. DFO and iron response in STAG2 KO cells.
(A, B, C) Effect of acute ∼16 h DFO treatment and 3 d recovery on (A) HAP1 WT and STAG2 KO cells, (B) H4 parent and STAG2 KI cells, and (C) 42 MGBA parent and STAG2 KI cells. (D) Western blots of HAP1 and 42 MGBA STAG2+/− cell lines after no treatment or treatment for ∼16 h of either 100 μM DFO or 300 μM ferric ammonium citrate.
- Figure 5. Knockout of cohesin loaders in HAP1 STAG2 KO cells.
(A, B, D) Clonogenic growth of HAP1 wild-type and STAG2 KO cells after transfection of sgRNAs for MAU2 (A), NIPBL (B) or PAGR1 (D). (C) Levels of cohesin proteins in lysates and soluble and chromatin fractions of HAP1 WT and STAG2 KO cells after transfection with control and MAU2 sgRNAs. *P < 0.05, **P < 0.005.
- Figure S5. Relationship between STAG2, STAG1, the cohesin loaders and PAXIP1/PAGR1.
(A, B, C) Effect of MAU2 CRISPR sgRNAs on growth of (A) RPE1 WT and STAG2 KO (B) H4 parent and STAG2 KI and (C) HAP1 WT and STAG1 KO cell lines. (D) Clonogenic assay after double transfection of pSpCas9-T2A plasmids marked with puro (for vector and sgSTAG2-1) and blast (for vector, sgMAU2-1 and sgSTAG1-1). Transfected cells were selected for both drugs for 3 d before being replated at single cell density for clonal counts. Grey dots represent the expected genetic interaction predicted by the two single gene KO’s using the multiplicative model. (E, F) Propidium iodide–stained cell cycle profile (E) and fraction of mitotic (pH3) cells (F) in HAP1 WT and STAG2 KO cells after MAU2 KO. (G) Gene dependency score of cohesin core (red), accessory (blue) and loader (green) genes in the DepMap 20Q1 CRISPR dataset. Data points are those outside the 1–99 percentile. (H) Correlation of the gene dependency scores of the two subunits of the cohesin loader complex across the DepMap cell lines in the dataset. (I) Correlation of gene dependency scores of all other genes with either NIPBL or MAU2. Both cohesin loader subunits show the same top three genes as being highly correlated across the DepMap cell line dataset. (J) Pearson correlation coefficients of select cohesin loader-related genes from the DepMap CRISPR dataset. ns- not significant, *P < 0.05, **P < 0.005.
Supplementary Materials
