Cis- and trans-regulations of pre-mRNA splicing by RNA editing enzymes influence cancer development

Abstract

RNA editing and splicing are the two major processes that dynamically regulate human transcriptome diversity. Despite growing evidence of crosstalk between RNA editing enzymes (mainly ADAR1) and splicing machineries, detailed mechanistic explanations and their biological importance in diseases, such as cancer are still lacking. Herein, we identify approximately a hundred high-confidence splicing events altered by ADAR1 and/or ADAR2, and ADAR1 or ADAR2 protein can regulate cassette exons in both directions. We unravel a binding tendency of ADARs to dsRNAs that involves GA-rich sequences for editing and splicing regulation. ADAR1 edits an intronic splicing silencer, leading to recruitment of SRSF7 and repression of exon inclusion. We also present a mechanism through which ADAR2 binds to dsRNA formed between GA-rich sequences and polypyrimidine (Py)-tract and precludes access of U2AF65 to 3' splice site. Furthermore, we find these ADARs-regulated splicing changes per se influence tumorigenesis, not merely byproducts of ADARs editing and binding.

Fig. 1. ADAR proteins regulate a subset of alternative splicing events.

a WB analyses of ADAR1 and ADAR2 proteins in EC109 cells that were stably knocked down (shADAR1 #3 and #9; shADAR2 #939 and #942; and scramble shRNA (scr)) or overexpressed (pLenti-ADAR1; pLenti-ADAR2; and empty vector control) for ADAR1 or 2, using lentiviral system. β-actin (actin) was used as a loading control. b Pie charts representing the number (and percentage) of each type of alternative splicing events affected by ADAR1 (left) and ADAR2 (right). c Heat maps showing the differentially spliced cassette exon events, upon knockdown and overexpression of ADAR1 (left) and ADAR2 (right). Splicing index (SI) is calculated by the ratio of inclusion junction reads to the sum of inclusion and skipping junction reads, and ΔSI indicates the difference in SI between ADARs knockdown/overexpression and their corresponding control samples. d, e RT-PCR analyses of representative ADAR1- d or ADAR2- e affected cassette exons in original RNA-Seq EC109 cells, as well as HEK293T cells. PSI, percent spliced in. Source data are provided as a Source Data file.

Fig. 2. ADARs repress exon inclusion editing dependently and independently.

a RT-PCR analysis of CCDC15-ex9 inclusion in HEK293T (left) and EC109 (right) cells that were transfected with the indicated wild-type ADAR1 (AR1) or ADAR2 (AR2) and different mutant (EAA and DeAD) forms of expression constructs (n = 4 biological replicates for each). EV, empty vector. b RT-PCR analysis of CCDC15-ex9 inclusion in HEK293T that were transfected with the indicated amount of EV, ADAR1, or ADAR2 construct (n = 3 biological replicates for each). Statistical significance of dose treatment is determined by linear regression. c RT-PCR analysis of RELL2-ex3 inclusion in the same samples as described in a (n = 3 or 4 biological replicates for each). a, c Data are presented as the mean ± S.D. of percent spliced in (PSI) values from biological replicates. Each dot represents a biological replicate. Statistical significance is determined by paired t-test (*P < 0.05; **P < 0.01). Source data are provided as a Source Data file.

Fig. 3. ADAR1-mediated editing of an ISS enhances SRSF7 binding for exon skipping.

a Sequence chromatograms illustrate the editing level of the indicated sites (1–4) at intron 8 of CCDC15 pre-mRNA in HEK293T cells that were transfected with empty vector control (EV), ADAR1 (0.25, 1.0, or 2.0 µg), or ADAR2 (2.0 µg) expression construct. Black arrowhead indicates editing position. Red arrows show the location of primers used for PCR amplification. b Upper panel: schematic diagram of wild-type (WT) CCDC15 exon 8–9–10 minigene. The positions where an A-to-G mutation was introduced are highlighted in red (sites 1, 2, and 4) and purple (site 3). The 13-bp region deleted in the Del minigene is shaded in orange. Lower panel: RT-PCR analysis of exon 9 inclusion of exogenous CCDC15 transcripts in HEK293T cells that were transfected with the indicated WT or mutant minigenes (n = 3 or 2 biological replicates for each). 1 + 2 denotes both sites 1 and 2 were mutated from A to G. c In silico prediction of SRSF7 binding sites on the edited CCDC15 pre-mRNA by Human Splicing Finder (orange line) and RBPmap (blue line). The edited nucleotide at site 2 is highlighted in red. d RT-PCR analysis of exon 9 inclusion of exogenous CCDC15 transcripts in HEK293T cells that were co-transfected with WT or site 2-mutated (Mut 2) minigene together with EV or SRSF7 expression construct (n = 3 biological replicates for each). e WB analysis of RNA pulldown products (eluate) shows binding of SRSF7 and hnRNPK protein to the WT or Mut 2 RNA probes. Sequence of the Mut 2 probe is shown in c and the WT probe is the same except site 2 that remains as an unedited adenosine. FT, flow-through. b, d Data are presented as the mean ± S.D. of percent spliced in (PSI) values from biological replicates. Each dot represents a biological replicate. Statistical significance is determined by paired t-test (*P < 0.05; **P < 0.01). Source data are provided as a Source Data file.

Fig. 4. Inter-intronic dsRNA is essential for ADAR1/2 binding and splicing regulation.

a Schematic diagram illustrates serial deletions introduced into intron 9. Red asterisk represents editing site. b RT-PCR analysis of exon 9 inclusion of exogenous CCDC15 transcripts in HEK293T cells that were co-transfected with the indicated minigene and overexpression construct (n = 3 or 2 biological replicates for each). Data are presented as the mean ± S.D. of percent spliced in (PSI) values from biological replicates. Each dot represents a biological replicate. Statistical significance is determined by paired t-test (*P < 0.05; **P < 0.01). c Sequence chromatograms illustrate the editing level of the indicated sites (1–4) in HEK293T cells that were co-transfected with the indicated minigene and overexpression construct. Black arrowhead indicates editing position. d In silico prediction of RNA secondary structure by RNAfold. Minimum free energy structures drawing encoding base-pair probabilities are shown. Base-pair probabilities are shown by a color spectrum. Arrow indicates the editing site. e REMSA analysis of binding of ADAR1 or ADAR2 protein to CCDC15 transcripts in vitro, using a ³²P-labeled RNA probe which simulates the dsRNA formed between introns 8 and 9 (CCDC15 In8-9 WT) together with the increasing amount of recombinant ADAR1/2 protein. f RIP-quantitative PCR (qPCR) analysis of the binding of ADAR1 or ADAR2 protein to exogenous CCDC15 transcripts (edited region in intron 8 and ECS in intron 9) in vivo (bottom panel). HEK293T cells were transfected with FLAG empty vector, FLAG-ADAR1, or FLAG-ADAR2, together with the wild-type CCDC15 minigene, followed by RIP assay at 48 h post transfection. WB analysis of FLAG-RIP immunoprecipitates is shown in the top panel. Input indicates 1% of the total cell lysate. Data is presented as mean ± S.D. of %input derived from qPCR technical triplicates from a representative experiment. Each dot represents a technical replicate. Statistical significance is determined by unpaired, two-tailed Student’s t-test (***P < 0.001). g REMSA analysis of the binding of SRSF7 to CCDC15 transcripts in the absence or presence of ADAR1 protein in vitro, using a ³²P-labeled CCDC15 In8-9 WT long dsRNA probe. Source data are provided as a Source Data file.

Fig. 5. ADAR2 binding to dsRNA that involves the Py-tract blocks U2AF65.

a Schematic diagram of RELL2 minigene and serial deletions in exon 3 and intron 3. Editing site is highlighted in red. b RT-PCR analysis of exon 3 inclusion of exogenous RELL2 transcripts in HEK293T cells that were co-transfected with the indicated minigene and overexpression construct (n = 2 biological replicates for each). c In silico prediction of RNA secondary structure by RNAfold. Base-pair probabilities are shown by a color spectrum. Asterisk indicates the editing site. d RT-PCR analysis of exon 3 inclusion of exogenous RELL2 transcripts in HEK293T cells that were co-transfected with the indicated minigene and overexpression construct (n = 2 biological replicates for each). GA-rich sequences #1 and #2 are highlighted in red and blue. e REMSA analysis of the binding of ADAR2 protein to RELL2 transcripts in vitro, using ³²P-labeled wild-type or mutant RELL2 Py-ex3 dsRNA probe with the increasing amount of recombinant ADAR2 protein. %Shift is calculated as shift band intensity over the sum of free probe and shift band intensities. f RIP-qPCR analysis of the binding of ADAR2 protein to exogenous RELL2 transcripts in vivo. WB analysis of FLAG-RIP immunoprecipitates is shown in the top panel. Input indicates 1% of the total cell lysate. Data are presented as mean ± S.D. of %input derived from qPCR technical triplicates from a representative experiment. Each dot represents a technical replicate. Statistical significance is determined by unpaired, two-tailed Student’s t-test (****P < 0.0001). g In vitro RNA–protein binding assay by UV crosslinking demonstrated the binding of U2AF65 to the RELL2 dsRNA probe, in the absence or presence of ADAR2 protein. h RNA pulldown assay detected the binding of U2AF65 protein to the dsRNA probe, in response to the addition of increasing amount of ADAR2 protein. WB analysis of U2AF65 and ADAR2 proteins in RNA pulldown (eluate) products and flow-through fractions. b, d Data are presented as the mean ± S.D. of percent spliced in (PSI) values from biological replicates. Each dot represents a biological replicate. Statistical significance is determined by paired t-test (*P < 0.05; **P < 0.01). Source data are provided as a Source Data file.

Fig. 6. Involvement of GA-rich motif(s) in ADARs-mediated splicing regulation.

a, b The top four motifs enriched in ADARs-regulated cassette exons a and non-ADARs-regulated cassette exons b as provided by MEME. Height of letter depicts occurrence of nucleotides at specific position. E-value represents statistical significance of the motif. Target represents the number of cassette exons containing a particular motif over total input number. c RNA secondary structure predictions for pre-mRNA sequences of representative ADARs-regulated cassette exons by RNAfold. Base-pair probabilities are shown by a color spectrum. Red line indicates the GA-rich motif. Asterisk denotes editing site. d Sequence chromatograms illustrate editing site adjacent to GA-rich motif of AKAP11 and KLRG1 upon ADAR2 overexpression in HEK293T cells. GA-rich motif is shaded in orange. Black arrowhead indicates editing position.

Fig. 7. Skipping of RELL2 exon 3 results in AS-NMD.

a Schematic diagram showing skipping of RELL2 exon 3 may trigger NMD. With exon 3 included, RELL2 pre-mRNA has a stop codon at the end of exon 6, which is then translated to a functional protein. In contrast, skipping of exon 3 results in a frameshift and creates a pre-mature termination codon at the beginning of exon 5, which locates 368 nt upstream of the last exon–exon junction. b RT-PCR analysis of different isoforms of the indicated transcripts in HEK293T cells upon inhibition of NMD. Cells were transfected with scramble shRNA (scr) or shRNA against UPF1 (shUPF1) for UPF1 knockdown, or treated with DMSO or CHX. SRSF1 or ZDHHC16 serves as a positive or negative control, respectively, to ensure successful inhibition of NMD. Number sign indicates the NMD-sensitive isoform. c qPCR analysis of total RELL2 transcript level in HEK293T cells with the indicated treatment. Data are presented as the mean ± S.D. of relative expression derived from qPCR technical triplicates or duplicates from a representative experiment. Each dot represents a technical replicate. Statistical significance is determined by unpaired, two-tailed Student’s t-test (*P < 0.05; **P < 0.01). d qPCR analysis of total RELL2 transcript level in HEK293T cells that were transfected with the increasing amount of EV (empty vector) or ADAR2 construct. Data are presented as the mean ± S.D. of relative expression derived from qPCR technical duplicates from a representative experiment. Each dot represents a technical replicate. Source data are provided as a Source Data file.

Fig. 8. ADARs-mediated splicing changes affect tumorigenesis.

a, b RT-PCR analysis of CCDC15-ex9 a or RELL2-ex3 b splicing pattern in 12 representative matched pairs of primary ESCC and NT samples. Patients demonstrating ≥1.5-fold higher expression of ADAR1 or/and ADAR2 in tumors than their matched NT tissues were classified as ADAR(1/2)^high group; while the remaining cases were included into ADAR(1 and 2)^normal/low group. Patients demonstrating ≥1.5-fold higher expression of ADAR2 in tumors than their matched NT tissues were classified as ADAR2^high group; while the remaining cases were included into ADAR2^normal/low group. Tumor samples with >5% increase in exon inclusion as compared to NT samples were defined as inclusion^up group; while tumor samples with >5% decrease in exon inclusion as compared to NT samples were defined as inclusion^down group. Red, blue, and gray lines indicate inclusion^up, inclusion^down, and no change cases, respectively. c, f, i Quantification of foci formation induced by the indicated cells (n = 2 or 3 biological replicates for each). Scale bar, 1 cm. d, g, j Quantification of colonies formed in soft agar induced by the indicated cells (n = 2 or 3 biological replicates for each). Scale bar, 200 µm. e, h, k Representative image of xenograft tumors derived from the indicated cells (n = 7 e, 6 h, or 5 k mice per group) at end point (left panel). Growth curves of tumors derived from the indicate cells are shown in right panel. c, d, f, g, i, j Data are presented as the mean ± S.D. of biological replicates from a representative experiment. Each dot represents a biological replicate. Statistical significance is determined by unpaired, two-tailed Student’s t-test (*P < 0.05; ****P < 0.0001). e, h, k Data are presented as the mean ± S.D. of tumor volumes. Statistical significance is determined by unpaired, two-tailed Student’s t-test (*P < 0.05; **P < 0.01). Source data are provided as a Source Data file.

Fig. 9. Schematic mechanistic diagram of ADARs-regulated alternative splicing.

Conversion of adenosine to inosine by ADAR1 at a GA-rich sequence proximal to cassette exon of a target transcript recruits SR proteins or other splicing factors to repress exon inclusion. Independent of their editing functions, ADAR1 and ADAR2 proteins can bind to inter-introns of a target transcript, thereby looping out the exon and blocking the access of the spliceosome to the splice site. Alternatively, ADAR2 blocks the access of U2AF65 to the 3′ splice site by binding to the dsRNA formed between the GA-rich sequence and Py-tract.