ARID1A deficiency modulates RNA editing levels and categories

To explore changes in gene enrichment resulting from ARID1A deletion, we performed Gene Oncology analysis on HCT116 ARID1A_WT and ARID1A_KO. Within the “Uniprot_keywords” category, we identified a significant upregulation of “RNA editing” in ARID1A_KO cell lines (Fig. 1A). Volcano plots were employed to depict the differentially expressed genes in the control and ARID1A knockdown group in HCT116 and A375 cell lines (Additional file 1: Fig. S1A-D).

Fig. 1

ARID1A deficiency modulates RNA editing levels and types of edited RNAs. A GO analysis showing “UP_keywords” category of HCT116_ARID1A_KO compared with ARID1A_WT cell lines. “UP_keywords” = “Uniprot_keywords”. n = 2. B Pie chart of percentage of A-I RNA editing sites contribution in HCT116_Scramble group. C Pie chart of percentage of A-I RNA editing sites contribution in HCT116_shARID1A group. D Pie chart of percentage of A-I RNA editing sites contribution in A375_Scramble group. E Pie chart of percentage of A-I RNA editing sites contribution in A375_shARID1A group. F The amount of A-I RNA editing sites and genes edited on exons between control and ARID1A knockdown A375 cell lines. G A-I RNA editing categories show the difference of shARID1A_log2 Ratio and Scramble_log2 Ratio between ARID1A knockdown and control groups in HCT116 cell line. H A-I RNA editing categories show the difference of shARID1A_log2 Ratio and Scramble_log2 Ratio between ARID1A knockdown and control groups in A375 cell line

A-I RNA editing manifests across diverse RNA types. In ARID1A_Scramble and shARID1A groups of HCT116, RNA editing predominantly occurred at short interspersed nuclear element (SINE), 3’UTR, and transcription termination sites (TTS) (Fig. 1B-C; Additional file 2: Table S1). In A375 cells, RNA editing was chiefly observed at short interspersed nuclear element (SINE), long interspersed nuclear element (LINE), and 3’UTR (Fig. 1D-E; Additional file 2: Table S2). ARID1A knockdown induced alterations in the distribution and percentage of RNA editing sites.

To further scrutinize the correlation between ARID1A deficiency and RNA editing, we examined the quantity of RNA editing sites in A375 cell lines. In cells with ARID1A deficiency, the number of editing sites increased by over 30,000 and the number of edited exons rose by 291 (Fig. 1F). Log2 ratio (obs / exp) differences between the ARID1A knockdown group and the control group indicated that, at highly edited sites such as SINE and LINE, no significant difference was observed. However, for low-edited sites like rRNA and scRNA, substantial differences existed between the two groups. Furthermore, variations in differential sites were noted between the HCT116 and A375 cell lines (Fig. 1G-H; Additional file 2: Tables S1 and S2).

Collectively, these findings suggest that ARID1A deficiency modulates both RNA editing levels and categories of RNA editing.

ARID1A deficiency changes RNA editing levels through ADAR1

Our findings indicated that ARID1A knockdown increases the number of RNA editing sites and induces changes in RNA editing categories. ADAR1, a pivotal RNA editor catalyzing A-I deamination, was examined for mRNA expression in HCT116 ARID1A knockdown and knockout cell lines. The mRNA levels of ADAR1 were also evaluated post ARID1A knockdown (Fig. 2A-B). In HCT116 ARID1A_KO cells, the overall RNA editing level increased, and ADAR1 knockdown in ARID1A_KO cells reversed this overall editing degree (Fig. 2C), suggesting that the alteration in RNA editing levels upon ARID1A loss is mediated by ADAR1. Next, we observed a significant increase in ADAR1 protein levels, particularly p150 rather than p110, following ARID1A depletion (Fig. 2D-G). Conversely, ADAR2 expression did not show an increase after ARID1A knockdown in HCT116, A375, and A2058 cell lines (Additional file 3: Fig. S2A).

Fig. 2

ARID1A deficiency upregulates overall RNA editing level mediated by ADAR1. A RT-qPCR results of ADAR1 mRNA expression level in HCT116_ARID1A_WT and HCT116_ARID1A_KO cell lines. n = 3; mean ± SD; *, P < 0.05. B RT-qPCR results of ARID1A and ADAR1 mRNA expression level in HCT116_Scramble, HCT116_shARID1A#1, and HCT116_shARID1A#2 cell lines. n = 3; mean ± SD; *, P < 0.05; ns, not significant. C RNA editing frequency of HCT116_ARID1A_WT, ARID1A_KO, and ARID1A_KO + shADAR1 cell lines. n = 2; mean ± SD; **, P < 0.01; ****, P < 0.0001. ns, not significant. D Western blot shows ARID1A and ADAR1 expression in HCT116_ARID1A_WT and HCT116_ARID1A_KO cell lines. E Data quantification of protein level of ARID1A, ADAR1 p150 and ADAR1 p110 in HCT116_ARID1A_WT and HCT116_ARID1A_KO cell lines. n = 3; mean ± SD; ****, P < 0.0001; ns, not significant. F Western blot shows ARID1A and ADAR1 expression in HCT116_Scramble, HCT116_shARID1A#1, and HCT116_shARID1A#2 cell lines. G Data quantification of protein level of ARID1A, ADAR1 p150 and ADAR1 p110 in HCT116_Scramble, HCT116_shARID1A#1, and HCT116_shARID1A#2 cell lines. n = 3; mean ± SD; ***, P < 0.001; ****, P < 0.0001; ns, not significant. H Western blot shows ADAR1 p150 and p110 levels in cytosolic and nuclear fractions in HCT116_ARID1A_WT and HCT116_ARID1A_KO. Cells were treated with IR 10 Gy and fixed at time point of 0 h, 1 h, 4 h, 16 h. I Western blot shows ADAR1 p150 and p110 levels in cytosolic and nuclear fractions in A375. J Western blot shows ADAR1 p150 and p110 levels in cytosolic and nuclear fractions in A 2058. K ADAR1 mRNA expression between ARID1A_WT and ARID1A_MUT in COAD patient specimens (TCGA database). The bar represents the mean ± SD. L ADAR1 mRNA expression between ARID1A_WT and ARID1A_MUT in STAD patient specimens (TCGA database). The bar represents the mean ± SD. M Kaplan–Meier curves showing TCGA UVM patient overall survival (OS) stratified by tumor ARID1A and ADAR1 mRNA expression. Red: ARID1A_high expression and ADAR_high expression; Green: ARID1A_high expression and ADAR_low expression; Blue: ARID1A_low expression and ADAR_high expression; Purple: ARID1A_low expression and ADAR_low expression

We further investigated whether ARID1A knockdown affected the subcellular distribution of ADAR1 in the nucleus and cytoplasm. Subcellular fractionation analysis revealed an increased cytoplasmic distribution of ADAR1 p150 after ARID1A knockout, while ADAR1 p110 remained concentrated in the nucleus (Fig. 2H). The distribution was unaffected by IR treatment and was consistent in A375 and A2058 (Fig. 2I-J).

Next, we examined the ADAR1 mRNA expression levels in TCGA database between ARID1A_WT and ARID1A_MUT groups. ADAR1 mRNA expression levels in colon adenocarcinoma (COAD) and stomach adenocarcinoma (STAD) patient samples were significantly higher in ARID1A_MUT samples (Fig. 2K-L). Kaplan–Meier curves revealed that UVM patients with low ARID1A expression and high ADAR1 expression exhibited the worst overall survival compared to those with high ARID1A expression, both ARID1A and ADAR1 high expression, and both low expression (Fig. 2M).

In summary, these data illustrate that ARID1A deficiency impacts altered RNA editing levels by upregulating ADAR1 and influences the cellular distribution of ADAR1 p150 and p110.

ARID1A deficiency slightly changes the editing level of the Q/R site of GlutR-B

Our data indicated that ARID1A deficiency leads to an increase in RNA editing levels. Consequently, we sought to investigate whether ARID1A deficiency could influence the RNA editing levels of specific genes. The Q/R site of the GlutR-B gene is among the most well-known RNA edited gene loci. Therefore, we examined whether the deletion of ARID1A resulted in an altered editing rate at this specific location. We designed primers for the GlutR-B Q/R site and amplified the specific fragment (Fig. 3A). The PCR products were subsequently cloned into a TA vector, and Sanger sequencing was performed on 100 colonies. The results of the edited and unedited GlutR-B Q/R sites were analyzed (Fig. 3B). ARID1A deficiency led to higher rates of Q/R site editing at the GlutR-B site in HCT116, A2058, and A375 cell lines (Fig. 3C–G). However, the RNA editing level of the GlutR-B Q/R site only exhibited a slight increase after ARID1A deficiency. This suggests that while ARID1A deficiency impacts the overall RNA editing level, it does not play a predominant role in the editing level of the GlutR-B Q/R site.

Fig. 3

ARID1A deficiency slightly changes the editing level of the Q/R site of GlutR-B. A Schematic overview of the TA cloning process. After PCR amplification of a segment of the gene containing the Q/R site of GlutR-B, the fragment was inserted into the T-vector. After ligation and transformation, 100 single clones were selected and plated for Sanger sequencing. Sequence alignment was performed using SnapGene software. B Comparison of GlutR-B Q/R site sequences in edited state and non-edited state. The red arrow indicates the Q/R site. C The Q/R site editing rate in HCT116 ARID1A_WT and ARID1A_KO cell lines. n = 3 of independent experiments; ***, P < 0.001. D Western blot shows ARID1A knockdown efficiency in A2058 shARID1A cells. E The Q/R site editing rate in HCT116 ARID1A_WT and ARID1A_KO cell lines. n = 3 of independent experiments; ns, not significant. F Western blot shows ARID1A knockdown efficiency in A375 shARID1A cells. G The Q/R site editing rate in HCT116 ARID1A_WT and ARID1A_KO cell lines. n = 3 of independent experiments; ns, not significant

The CDK13 gene can be edited by ADAR1 and ARID1A deficiency increased sensitivity to SR-4835

We proceeded to investigate whether other genes were subject to increase editing by ADAR1. Our bioinformatic analysis revealed an overall increase in RNA editing rates and modifications in the types of edited RNAs in cells with ARID1A knockdown. Moreover, the “RNA editing” category exhibited differences between the ARID1A knockdown and control groups. Genes within the RNA editing category, including ARL6IP4, BLCAP, COPA, CDK13, and NEIL1, were found to be altered by ARID1A knockdown, suggesting changes in the RNA editing levels of these genes after ARID1A deficiency (Fig. 4A). Consequently, we compared the RNA editing rates of genes in the ARID1A_WT, ARID1A_KO, ARID1A_KO + shADAR#1, and ARID1A_KO + shADAR#2 groups. The editing rate in COPA, BLCAP, and ARL6IP4 did not increase after ARID1A KO and decreased after ADAR1 knockdown (Fig. 4B).

Fig. 4

The CDK13 gene is edited by ADAR1 and ARID1A deficiency increased sensitivity to SR4835. A Comparison of HCT116 shARID1A to ARID1A_Scramble cell lines. The differential genes in “RNA editing” category consists of ARL6IP4, BLCAP, COPA, CDK13, and NEIL1. B The RNA editing frequency at different sites of COPA, BLCAP, ARL6IP4 genes in HCT116 ARID1A_WT, ARID1A_KO, ARID1A_KO + shADAR#1, and ARID1A_KO + shADAR#2 cell lines. C The RNA editing frequency at three different sites of CDK13 genes in HCT116 ARID1A_WT, ARID1A_KO, ARID1A_KO + shADAR#1, and ARID1A_KO + shADAR#2 cell lines. D RT-qPCR results show ADAR mRNA expression in HCT116 ARID1A_KO, ARID1A_KO + shADAR#1, and ARID1A_KO + shADAR#2 cell lines. n = 3; mean ± SD; *, P < 0.05; ns, not significant. E Diagram of the CDK13 kinase, showing the serine/threonine protein kinase active region. Intron regions and three RNA editing sites (K21R, K117R, and Q113R) are marked. K21R:7_39990302; K117R:7_39990590; Q113R: 7_39990578. Exported by "SMART" software, the red line represents intron Phase 2 and the blue line represents intron Phase 1. F Representative images of A375 Scramble and shARID1A cells treated with SR-4835 at a concentration of 1 nM and 10 nM. G Quantitative results represent the mean ± SD of three independent experiments. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. H Representative images of HCT116 Scramble and shARID1A cells treated with SR-4835 at a concentration of 1 nM and 10 nM. I Quantitative results represent the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001

Results indicated a simultaneous increase in the RNA editing rate of the CDK13 gene at three loci, 7_39990302_CDK13_ + _Nonsyn_Lys- > Arg (K21R), 7_39990578_CDK13_ + _ Nonsyn_Gln- > Arg (Q113R), and 7_39990590_CDK13_ + _Nonsyn_Lys- > Arg (K117R) after ARID1A knockout, signifying a substantial rise in RNA editing of CDK13 gene following ARID1A deletion. Upon the combination of ADAR1 knockdown, the RNA editing level of CDK13 mediated by ARID1A deletion decreased, indicating that the change in RNA editing level of CD13 is achieved through ADAR1 (Fig. 4C). The knockdown efficiency of ADAR1 was verified by RT-qPCR (Fig. 4D). The three editing sites were marked on the CDK13 protein, all of which were distant from catalytic domain (Fig. 4E).

Considering that CDK13 plays a crucial role in cell cycle and participates in RNA editing, we selected CDK13 for further investigation. ARID1A deletion enhanced RNA editing of CDK13, potentially compromising its biological activity. We then explored whether the CDK13 inhibitor could be employed to assess the role of CDK13 and whether the therapeutic effect was influenced by ARID1A. Various CDK13 inhibitors, including SR-4835, highly selective for CDK12 and CDK13 with an ATP-competitive binding mechanism, were considered. It interacts with the kinase’s hinge region by hydrogen bonding [19]. SR-4835 exhibits anticancer activity in ovarian cancer, triple-negative breast cancer, and leukemia [19,20,21]. In the absence of pharmacological treatment, ARID1A knockdown markedly decreased tumor clone development; under SR-4835 treatment, ARID1A-deficient cells demonstrated improved sensitivity to CDK13 inhibitors (Fig. 4F-I; Additional file 4: Fig. S3A-C).

The Q113R and K117R of CDK13 are edited by ADAR1 and ADAR1 knockdown rescued sensitivity to SR-4835

In the subsequent experiments, we conducted RT-PCR and Sanger sequencing to validate the RNA editing status at the K21R, Q113R, and K117R sites, as identified in the bioinformatics analysis. No corresponding edits were detected at the K21R site (Fig. 5A). Notably, at the Q113R and K117R positions, A-to-G editing events were observed following ARID1A knockout (Fig. 5B-C).

Fig. 5

ADAR1 edits CDK13 at position Q113R and K117R, ADAR1 knockdown rescues the sensitivity to SR4835. A Sanger sequencing chromatograms using a reverse primer illustrate editing of CDK13 K21R position in ARID1A_WT, ARID1A_KO and genomic DNA (gDNA) in HCT116 cell lines. B Sanger sequencing chromatograms using a reverse primer illustrate editing of CDK13 Q113R position in ARID1A_WT, ARID1A_KO and genomic DNA (gDNA) in HCT116 cell lines. C Sanger sequencing chromatograms using a reverse primer illustrate editing of CDK13 K117R position in ARID1A_WT, ARID1A_KO and genomic DNA (gDNA) in HCT116 cell lines. D Representative images of HCT116_ARID1A_KO cell lines transfected with siNC and siADAR and treatment with SR-4835 at a concentration of 5 nM and 10 nM. E Quantitative results represent the mean ± SD of three independent experiments. **, P < 0.01; ***, P < 0.001. F Representative images of A375_ARID1A_KO cell lines transfected with siNC and siADAR and treatment with SR-4835 at a concentration of 5 nM and 10 nM. G Quantitative results represent the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01. H Representative images of DLD-1_ARID1A_KO cell lines transfected with siNC and siADAR and treatment with SR-4835 at a concentration of 5 nM and 10 nM. I Quantitative results represent the mean ± SD of three independent experiments. ***, P < 0.001.****, P < 0.0001

Furthermore, when ARID1A KO was combined with siNC, cells displayed heightened sensitivity to SR4835. Conversely, when combined with ADAR1 knockdown, cells exhibited reduced sensitivity to SR4835 treatment across HCT116, A375, and DLD-1 cell lines (Fig. 5D-I). This not only supports the observed correlation between ARID1A’s interaction with ADAR but also underscores its impact on the effectiveness of the CDK13 inhibitor.

ADAR1 deficiency results in altered cell cycle in response to ionizing radiation (IR)

Next, we investigated the influence of ADAR1 on the cell cycle following ARID1A deletion. First, we assessed the cell cycle distribution of ARID1A_WT and ARID1A_KO cells at different time points post-IR. At 4 and 8 h after irradiation, both ARID1A_WT and ARID1A_KO cells exhibited a significant increase in the G2-M population (Additional file 4: Fig. S3D; Additional file 5: Fig. S4A–B). At 16 h post irradiation, a substantial proportion of ARID1A_WT cells remained arrested at the G2-M checkpoint, while the G2M checkpoint arrest in ARID1A-depleted cells was notably reduced. Simultaneously, there was a significant increase in the proportion of cells in the G1 phase.

Upon introducing ADAR1 knockdown in ARID1A_WT cells, we observed a decrease in the proportion of cells in the G1 phase and a more rapid entry into G2-M phases at 4 and 8 h post IR. However, this altered distribution disappeared when ARID1A knockout was simultaneously applied, suggesting that ADAR1 deletion induces changes in the cell cycle in ARID1A-deficient cells. Additionally, ADAR1 knockdown resulted in an increased G2-M arrest in the context of ARID1A-KO.

To further assess the impact on mitotic cells, we utilized phospho-Histone H3 staining. In ADAR1-knockdown ARID1A_KO cells exposed to IR, we observed an increase in cells re-entering to M phase, especially under Taxol treatment. However, this increase was not observed following combined ARID1A deficiency (Additional file 5: Fig. S4C). Collectively, these results suggest that ADAR1 plays a role in cell cycle distribution, indicating potential implications for RNA editing of CDK13. Notably, knockdown of CDK13 in HCT116_ARID1A_WT cells led to a decrease in the G1 phase and an increase in the G2-M phase, which was enhanced upon concomitant ARID1A KO (Additional file 4: Fig. S3E; Additional file 5: Fig. S4D-E).