Incomplete paralog compensation generates selective dependency on TRA2A in cancer

Abstract

Paralogs often exhibit functional redundancy, allowing them to effectively compensate for each other's loss. However, this buffering mechanism is frequently disrupted in cancer, exposing unique paralog-specific vulnerabilities. Here, we identify a selective dependency on the splicing factor TRA2A. We find that TRA2A and its paralog TRA2B are synthetic lethal partners that function as widespread and largely redundant activators of both alternative and constitutive splicing. While loss of TRA2A alone is typically neutral due to compensation by TRA2B, we discover that a subset of cancer cell lines are highly TRA2A-dependent. Upon TRA2A depletion, these cell lines exhibit a lack of paralog buffering specifically on shared splicing targets, leading to defects in mitosis and cell death. Notably, TRA2B overexpression rescues both the aberrant splicing and lethality associated with TRA2A loss, indicating that paralog compensation is dosage-sensitive. Together, these findings reveal a complex dosage-dependent relationship between paralogous splicing factors, and highlight how dysfunctional paralog buffering can create a selective dependency in cancer.

Fig 1. TRA2A is a selective dependency in a subset of cancer cell lines.

(A) Distribution of gene effect scores after TRA2A knockout from the Cancer Dependency Map (Public 24Q2 dataset). Red lines mark gene effect scores of -0.5 (dotted) and -1.0 (solid). (B) Immunoblot showing depletion of TRA2A protein levels after CRISPR-Cas9 targeting in NCI-H23 cells. ACTB was used as a loading control. (C,D) Competition-based proliferation assays performed in indicated Cas9 + cell lines after TRA2A KO. SgRNA+ populations were monitored over time with a co-expressed fluorescent protein marker. Plotted is the relative sgRNA+ population normalized to the Day 3 sgRNA+ population. Bar graph summarizes normalized percentage of sgRNA+ cells remaining at Day 22 or 23 after infection, n = 3. (E) Immunoblot showing depletion or rescue of TRA2A protein levels in NCI-H23 Cas9 + cells stably expressing luciferase (control) or TRA2A cDNA resistant to sg4 (sg4-res) and infected with control or TRA2A-targeting sgRNAs. (F) Competition-based rescue assay performed in indicated TRA2A-dependent Cas9 + cells also stably expressing luciferase or TRA2A cDNA resistant to sg4 (sg4-res) and subsequently infected with control or TRA2A-targeting sgRNAs. Shown is the percentage of sgRNA+ cells at Day 28 after infection, normalized to Day 3, n = 3. (G) Immunoblot showing depletion of TRA2A protein levels after CRISPRi-mediated repression in NCI-H23 cells. (H) Competition-based proliferation assays performed in indicated Zim3-dCas9 + cell lines after TRA2A KD. SgRNA+ populations were monitored over time with a co-expressed fluorescent protein marker. Shown is the percentage of sgRNA+ cells at Day 28 after infection normalized to Day 3, n = 3. (I) Immunoblots showing protein levels of TRA2A, TRA2B, and ACTB in TRA2A-independent and -dependent cell lines. Quantification was performed by normalizing protein signal to total protein stain (see S1D Fig), n = 2. (J) Summary of top 10 TRA2A-dependent cell lines and 3 tumor type-matched TRA2A-independent cell lines from DepMap. (D,F,H,I) Error bars represent standard deviation from the mean. (**)P < 0.01, (***)P < 0.001, (****)P < 0.0001, and (ns) not significant, as calculated by repeated measures two-way ANOVA followed by Dunnett’s multiple comparison test.

Fig 2. Genetic modifier screening identifies paralog synthetic lethality between TRA2A and TRA2B.

(A) Schematic representation of CRISPR screen for genetic modifiers of TRA2A dependency. (B) Scatterplot of log2 fold-change (logFC) in A549 cells between control compared to the plasmid DNA library (x-axis) and TRA2A KO compared to the plasmid DNA library (y-axis), calculated by MAGeCK. (C) (Left) Distribution of genes ranked by the log2 fold-change (logFC) between TRA2A KO and control conditions in A549, calculated by MAGeCK. The blue and red dots indicate two standard deviations from the mean log2 fold-change. (Right) Log2 fold-change between TRA2A KO and control conditions of the individual guides targeting genes representing the 10 most depleted genes in A549 cells. (D) Log2 fold-change of TRA2B sgRNAs between control or TRA2A KO conditions relative to plasmid DNA library. (E) Immunoblot confirming dual depletion of TRA2A and TRA2B protein levels after CRISPRi targeting. ACTB was used as loading control. (F) Dual-guide CRISPRi competition assays in indicated Zim3-dCas9 + cell lines, n = 3. Control condition indicates sgRNAs targeting AAVS1 and ROSA26 safe harbor loci, AARS1 KD indicates pan-essential positive control. Values indicate Day 28 percent sgRNA+ population normalized to Day 3 percent sgRNA+ population. Error bars represent standard deviation from the mean. (****)P < 0.0001, as calculated by repeated measures two-way ANOVA followed by Dunnett’s multiple comparison test. (G) Quantification of percent AnnexinV+ apoptotic cells upon negative control or TRA2A/B DKD at Day 7 post-infection as measured by flow cytometry in the indicated cell lines. Error bars represent standard deviation from the mean. (*)P < 0.05, (**)P < 0.01, (***)P < 0.001, and (****)P < 0.0001, as calculated by two-tailed unpaired t-test. Schematic in (A) created in BioRender under License: https://BioRender.com/o10ks6p.

Fig 3. TRA2A and TRA2B function redundantly to maintain widespread constitutive splicing.

(A,B) Quantification of number of changing LSVs in (A) A549 and (B) NCI-H23 cells. Changing LSVs were determined by confidence level (probability changing) > 95% and |dPSI| > 10%. Blue represents negative dPSI or LSVs more excluded/repressed upon knockdown. Red represents positive dPSI or LSVs more included/activated upon knockdown. (C,D) Distribution of starting PSI value of changing LSVs in each knockdown condition compared to control in (C) A549 and (D) NCI-H23 cells. Red dashed line marks PSI = 95%. (E,F) Mean PSI level for all changing LSVs in control cells and after each knockdown, or subsetted for changing LSVs starting with PSI > 95% in control cells before each knockdown in (E) A549 and (F) NCI-H23 cells. (G) UpSet plot representing the overlap of changing LSVs upon TRA2A/B DKD in A549 and NCI-H23 cells. (H) Top enriched Reactome pathways of changing LSVs upon TRA2A/B DKD in A549 and NCI-H23 cells. (I) Scatterplot showing enrichment of Reactome pathways for changing LSVs upon TRA2A/B DKD in A549 and NCI-H23 cells. Data plotted represents the -log(adjusted p-value) assigned to the Reactome term.

Fig 4. TRA2A dependency is associated with lack of paralog compensation in splicing of cell cycle-related genes.

(A) UpSet plot representing the overlap of changing LSVs upon TRA2A KD in NCI-H23 and A549 cells. (B) Scatterplot showing enrichment of Reactome pathways for changing LSVs upon TRA2A depletion in A549 and NCI-H23. Data plotted represents the -log(adjusted p-value) assigned to the Reactome term. (C) Top enriched Reactome pathways of changing LSVs responsive to TRA2A KD in NCI-H23 cells. (D) (Top) Mean dPSI values of changing LSVs repressed upon TRA2A KD in NCI-H23 cells, plotted for both NCI-H23 and A549 cells. (Bottom) Heatmap representation of the same LSVs showing their mean PSI values across all conditions in NCI-H23 and A549. Rows represent an individual LSV and heat color represents the mean PSI for a given LSV, n = 3. (E) (Top) Mean dPSI values of changing LSVs repressed upon TRA2A KD in A549 cells, plotted for both A549 and NCI-H23 cells. (Bottom) Heatmap representation of the same LSVs showing their mean PSI values across all conditions in A549 and NCI-H23. Rows represent an individual LSV and heat color represents the mean PSI for a given LSV, n = 3. (F) RT-PCR validation of selected LSVs in OFD1 and CHEK1, performed in indicated TRA2A-dependent and independent cell lines.

Fig 5. Loss of TRA2A in TRA2A-dependent cells results in cell death from defects in mitosis.

(A) (Left) Percentage of cells in each cell cycle stage upon control or TRA2A KD in NCI-H23 cells at Day 7 after infection, measured by propidium iodide and phospho-Histone H3 staining followed by flow cytometry, n = 3. Error bars represent standard deviation from the mean. (*)P < 0.05, (**)P < 0.01, (***)P < 0.001, (****)P < 0.0001, and (ns) not significant, as calculated by repeated measures two-way ANOVA followed by Dunnett’s multiple comparison test. (Right) Representative flow cytometry histogram of propidium iodide staining showing cell cycle distribution upon control and TRA2A KD in NCI-H23 cells. (B) Percentage of apoptotic cells upon control or TRA2A KD in NCI-H23 at the indicated times after infection, measured by staining for Annexin-V followed by flow cytometry, n = 3. Error bars represent standard deviation from the mean. (**)P < 0.01, (***)P < 0.001, and (****)P < 0.0001, as calculated by repeated measures two-way ANOVA followed by Dunnett’s multiple comparison test. (C) Length of mitosis upon control or TRA2A KD in NCI-H23 cells, measured by live imaging, n = 3. Each dot represents a cell showing completion of mitosis (control n = 123, TRA2A KD n = 209). (ns) not significant, as calculated by two-tailed unpaired t-test. (D) Percentage of mitotic cell population undergoing death during mitosis upon control or TRA2A KD in NCI-H23 cells, measured by live cell imaging, n = 3. Error bars represent standard deviation from the mean. (**)P < 0.01, as calculated by two-tailed unpaired t-test. (E) Representative images of cells undergoing mitotic death upon control or TRA2A KD in NCI-H23 cells stably expressing a chromatin marker (H2B-mNeonGreen; green) and stained with an alpha tubulin dye (magenta). Scale bars = 10 um. For full movies, see S1 Movie.

Fig 6. TRA2A dependency is rescued by overexpression of TRA2B.

(A) Immunoblot and quantification of TRA2B protein expression relative to ACTB in NCI-H23 and LN319 cells stably expressing luciferase control (Luc) or overexpressing TRA2B (TRA2B-OE), n = 3. Error bars represent standard deviation from the mean. (**)P < 0.01 and (***)P < 0.001, as calculated by two-tailed unpaired t-test. (B) Competition assay performed in NCI-H23 and LN319 Cas9 + cells stably expressing luciferase control (Luc) or overexpressing TRA2B (TRA2B-OE) and subsequently infected with control or TRA2A sgRNAs, n = 3. Error bars represent standard deviation from the mean. (***)P < 0.001, (****)P < 0.0001, and (ns) not significant, as calculated by repeated measures two-way ANOVA followed by Tukey’s multiple comparison test. (C) Hierarchical clustering of splicing changes detected by MAJIQ in LN319 cells stably expressing luciferase control (Luc) or overexpressing TRA2B (TRA2B-OE) and then infected with control or TRA2A sgRNAs (TRA2A KO). Rows represent LSVs responsive to TRA2A KO compared to control from LN319 cells expressing luciferase. Heat represents the mean PSI for a given LSV, n = 3. (D) Mean dPSI of TRA2A KO-responsive LSVs for clusters C1 and C2 in luciferase and TRA2B-OE conditions. (E) RT-PCR validation of candidate TRA2A KO-responsive splicing changes that are rescued by TRA2B overexpression in NCI-H23 and LN319 cells. (F) Proposed model of TRA2A/TRA2B compensatory mechanisms in normal cells (left), upon TRA2A loss in cells with an intact paralog buffering mechanism (middle), and upon TRA2A loss in cells with a compromised paralog buffering mechanism (right).