MDM2 is an oncogene and critical negative regulator of tumor suppressor p53. Genotoxic stress causes alternative splicing of MDM2 transcripts, which leads to alterations in p53 activity and contributes to tumorigenesis. MDM2-ALT1 is one of the alternatively spliced transcripts predominantly produced in response to genotoxic stress, and is comprised of terminal coding exons 3 and 12. Previously, we found that SRSF1 induces MDM2-ALT1 by promoting MDM2 exon 11 skipping. Here we report that splicing regulator SRSF2 antagonizes the regulation of SRSF1 by facilitating the inclusion of exon 11 through binding at two conserved exonic splicing enhancers. Overexpression of SRSF2 reduced the generation of MDM2-ALT1 under genotoxic stress, whereas SRSF2 knockdown induced the expression of MDM2-ALT1 in the absence of genotoxic stress. Blocking the exon 11 SRSF2-binding sites using oligonucleotides promoted MDM2-ALT1 splicing and induced p53 protein expression, and apoptosis in p53 wild-type cells. The regulation of MDM2 splicing by SRSF2 is also conserved in mice, as mutation of one SRSF2-binding site in Mdm2 exon 11, using CRISPR-Cas9, increased the expression of the MDM2-ALT1 homolog Mdm2-MS2.
Taken together, the data indicate that modulating MDM2 splicing may be a useful tool for fine-tuning p53 activity in response to genotoxic stress.
This article is featured in Highlights of This Issue, p. 183
Murine Double Minute 2 (MDM2) is a proto-oncogene and critical negative regulator of the p53 tumor suppressor protein. MDM2 is overexpressed in many types of cancer, including osteosarcoma and other soft tissue sarcomas (1, 2). In response to genotoxic stress, MDM2 undergoes alternative splicing to generate splice variants that are unable to regulate p53 expression (3–5). This results in the stabilization of p53 and upregulation of pathways involved in apoptosis and cell-cycle arrest (6–8). Current therapies to inhibit the interaction between MDM2/p53 in p53 wild-type cancers have had limited success due to toxicity and because they do not also inhibit the regulation of p53 by MDMX. We have previously shown that an isoform of MDM2 containing exons 3 and 12, MDM2-ALT1 (MDM2-B), can bind and inhibit the proper localization of both full-length MDM2 and MDMX (6). Therefore, understanding the regulation governing this alternative splice variant presents an avenue to stabilize p53 in these cancers for therapeutic benefit.
One of the largest families of splicing regulatory proteins is the SR (Serine-Arginine) family of proteins. The role of SR proteins is to enhance or repress the recognition of exons, allowing for increased protein diversity (9–12). Alternative splicing changes have been associated with multiple cancer cell hallmarks and contribute to tumor progression and therapeutic resistance (13). However, the interplay of protein family members that regulate alternative splicing and contribute to the oncogenic transformation is not well understood.
To study the alternative splicing of MDM2 we have developed a damage-inducible minigene system. The MDM2 3-11-12s minigene recapitulates the splicing of the endogenous gene by excluding its intervening exon under UVC and cisplatinum stress (14, 15). We have previously identified one SR protein, SRSF1 (ASF/SF2), as a negative regulator of MDM2 splicing that supports the formation of the MDM2-ALT1 isoform in response to genotoxic stress (15). We have also identified FUBP1 as another RNA-binding protein that facilitates the full-length splicing of MDM2 (14, 16). However, our MDM2 3-11-12s minigene, which lacks the FUBP1-binding site, is still efficiently spliced under normal conditions. Therefore, we hypothesize that there are additional sites capable of regulating MDM2 alternative splicing. Here we report the identification of one such protein, SRSF2 (SC35). We show that mutation of the SRSF2-binding sites or targeting them with splice-switching oligonucleotides (SSO) increases the expression of the alternatively-spliced transcript MDM2-ALT1 and p53 protein expression.
Materials and Methods
Cell culture, growth, and transfection conditions
MCF7, NIH 3T3 cells, HeLa, and C2C12 cells were obtained from ATCC and Rh18/SMS-CTR cells were obtained from Peter J. Houghton (Greehey Children's Cancer Research Institute, San Antonio, TX). HeLa S3 cells were obtained from Hua Lou (Case Western University, Cleveland, OH). MCF7 and HeLa cells were chosen because these cells lines have been studied extensively from the standpoint of MDM2 alternative splicing and because they are robust in their tolerance to genotoxic stress, as well as amenable to genetic manipulation by transfection. SMS-CTR and Rh18 cells were chosen because they are rhabdomyosarcoma cell lines that do not demonstrate MDM2 alternative splicing. All human cell lines have been verified by short tandem repeat analysis (Genetica). Cells were examined for Mycoplasma contamination by DAPI staining. Experiments were performed within the first 10 passages of thawing cells. MCF7, NIH 3T3, HeLa, and C2C12 cell lines were maintained in DMEM, whereas SMS-CTR, Rh18, and HeLa S3 cells were maintained in RPMI medium. All were supplemented with 10% FBS (Thermo Fisher Scientific), 1× l-glutamine (Corning), and 1× penicillin/streptomycin (Corning). MCF7 cells were transfected as described previously (15). For RNA immunoprecipitation assays, MCF7 cells were transfected with T7-SRSF2 for 24 hours, treated under normal conditions or subjected to 50 J/m2 UVC, and harvested 24 hours later.
Plasmids and protein expression constructs
The LacZ plasmid has been described previously (15). The T7-SRSF2 construct was provided by Dr. Adrian Krainer (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). The MDM2 3-11-12s minigene was described previously (15). The SpCas9-2A-EGFP plasmid was obtained from Addgene. The guide sequences for target sequences were cloned into the BbsI site as described previously (17). The Mdm2 3-11-12s and Mdm2 4-11-12s minigene was constructed using PCR. PCR products were visualized under long-wave ultraviolet light, excised, and gel purified using QIAquick Gel Extraction Kit. A final multiplex PCR was performed with the two purified PCR products and two terminal primers. See Supplementary Table S1. The Mdm2 3-11-12s minigene was then cloned into the EcoRI-XhoI sites of the pCMV-Tag2B vector using T4 DNA ligase (NEB) according to the manufacturer's instructions.
RT and PCRs
Reverse transcription (RT) reactions were carried out using Transcriptor RT enzyme (Roche Diagnostics). Nonquantitative endogenous MDM2 PCRs were performed as reported previously (18). MDM2 minigene PCRs were performed as reported previously (14). Mdm2 amplicons and PCRs for MDM2 after RNA immunoprecipitation of T7-SRSF2 were performed using a set of nested primers under standard PCR conditions using Taq polymerase (Sigma).
Percentages of full-length and exon-excluded products were quantitated using ImageQuant TL (Version 8.1). Significance of all results were assessed using the two-tailed Student t test using GraphPad Prism (Version 8.0) unless otherwise noted.
Western blot analysis and antibodies
Cells were lysed in NP-40 buffer and equal amounts of protein were loaded in SDS sample buffer onto a SDS-PAGE gel, blotted onto a polyvinylidene difluoride membrane, and analyzed for expression of SRSF2 1SC-4F11 (EMD Millipore), T7-Tag (EMD Millipore), β-Galactosidase (Promega), recombinant Lamin A/C (Abcam), Nucleolin C23 (Sigma), GAPDH 14C10 (Cell Signaling Technology), β-Tubulin E7 (DSHB), or β-Actin AC-15 (Sigma). Protein sizes were determined using the Precision Plus Protein Dual Color Standards marker (Life Technologies).
Cells were seeded on coverslips and either treated under normal conditions or subjected to 50 J/m2 UVC. After 12 hours cells were fixed in 4% paraformaldehyde and permeabilized in 0.25% Triton X-100. Cells were then blocked in 10% donkey serum and incubated in primary antibodies SRSF2 αSC35 (BD Biosciences), T7-Tag (EMD Millipore), or SRRM2 (Atlas) in 5% donkey serum at 4°C overnight. Cells were incubated in secondary antibody at room temperature for 1 hour in dark (anti-mIgG Alexa Fluor 488 or anti-rIgG Alexa Fluor 647, Thermo Fisher Scientific). Coverslips were then mounted with 1 drop of Diamond ProLong Antifade with DAPI (Thermo Fisher Scientific) and cured overnight at room temperature. Cells were then imaged on a confocal microscope.
Cells were scraped from adherent plates in 1 mL of PBS on ice. Suspensions were transferred to Eppendorf tubes and spun down for 30 seconds at 1,300 × g. PBS was aspirated and cells were lysed in polysome lysis buffer (100 mmol/L KCl, 5 mmol/L MgCl2, 10 mmol/L HEPES, pH 7.0, 0.5% Nonidet P-40, 100 U/mL RNase inhibitor, Halt protease inhibitor) on ice for 5 minutes. Lysates were centrifuged at 14,000 × g for 15 minutes at 4°C and supernatant was transferred to a fresh tube. Approximately 1.5 mg of protein lysate was immunoprecipitated in a 1 mL reaction containing NT2 buffer (50 mmol/L Tris, pH 7.4, 150 mmol/L NaCl, 1 mmol/L MgCl2, 0.05% Nonidet p-40), 20 μg of T7-SRSF1, or mIgG isotype control with 15 mmol/L EDTA, pH 8.0, for 10 minutes at room temperature. One-hundred microliters of prewashed Dynal Protein G magnetic beads (Thermo Fisher Scientific) were then added to the immunoprecipitation reaction for an additional 10 minutes. Immunoprecipitates (IPs) were washed three times with 500 μL NT2 buffer, then twice with 500 μL PBS (containing 100 U/mL RNase inhibitor). IPs were resuspended in 150 μL proteinase K buffer (1.2 mg/mL proteinase K, 1% SDS, 100 U/mL RNase inhibitor in NT2 buffer) and incubated 30 minutes at 55°C. Beads were immobilized and supernatant was transferred to fresh Eppendorf tubes, to which 350 μL buffer RLT was added to both IP and 1/100 input samples. A Qiagen RNeasy protocol with DNase I digestion was then performed according to manufacturer's instructions.
RNA oligonucleotide pull down assay
RNA (5 nmol) was modified, conjugated to Adipic acid dihydrazide agarose beads (Sigma), and washed as described previously (15). RNA was then incubated in a splicing reaction at 30°C for 40 minutes, gently mixing every 5 minutes. Protein-bound beads were washed three times in Buffer D, then eluted in 40 μL 2X SDS buffer. Beads were boiled at 100°C for 5 minutes, and then spun down at 10,000 rpm at 4°C for 10 minutes. Eluates were collected and loaded in on SDS-PAGE gel as described above.
The siRNAs targeting human SRSF2 or a nonspecific siRNA were transfected into MCF7 cells at a concentration of 30 nmol/L, mediated by Lipofectamine RNAiMAX (Thermo Fisher Scientific) for a total of 72 hours. Posttransfection, cells were harvested for total RNA using an RNeasy kit (Qiagen) and subject to RT-PCR as described above. Protein was also collected as described above to confirm knockdown of SRSF2.
2′O-methyl SSOs specific to MDM2 exon 11 or a nonspecific SSO were transfected in MCF7 and Rh18 cells with either Lipofectamine 2000 (Thermo Fisher Scientific) for 24 hours (MCF7) or 4 hours (Rh18). SMS-CTR cells were nucleofected with 250 nmol/L SSOs using Nucleofector Kit R (Lonza) with program X-001 on an Amaxa Nucleofector II device. Cells were harvested for RNA after transfection using an RNeasy kit (Qiagen) and subjected to qPCR using conditions described below.
Quantitative real-time PCR
All Quantitative real-time PCR (qPCR) was performed with standard PCR conditions for using an Applied Biosystems 7900HT Fast Real Time PCR system (Life Technologies). Real-time PCR reactions were carried out using the SYBR Green PCR Master Mix (Applied Biosystems). The primers used to amplify the p53-target transcripts and MDM2-ALT1 have been described previously (6). All PCR reactions were carried out with three technical replicates and the amplification of single PCR products in each reaction was confirmed using dissociation curve.
CRISPR-Cas9 genome editing
NIH 3T3 cells were transfected with either a control plasmid SpCas9-2A-EGFP or SpCas9-2A-EGFP-MDM2 along with a HDR custom single-stranded DNA 243 base pair repair template. Four hours after transfection, cells were treated with a 1 mmol/L SCR7 (Xcessbio). Forty-eight hours after transfection, cells were sorted for GFP expression on a BD Influx FACS cell sorter running SortWare software. GFP-positive cells were plated and maintained in medium containing 1 mmol/L SCR7. After two passages, cells were collected for genomic DNA. Genomic changes were verified by Sanger sequencing.
Live cellular growth assay
Growth curves were performed with triplicate plating of either NIH 3T3 Control CRISPR, NIH 3T3 G165T CRISPR1, or NIH 3T3 G165T CRISPR2 cell lines. Cells were seeded at a density of 8 × 104 cells per well in a 12-well plate and analyzed for confluency using the IncuCyte ZOOM live cell imaging system taking pictures every 4 hours. Statistical significance was calculated by two-way ANOVA.
Mutations in SRSF2-binding sites cause exon exclusion in the MDM2 3-11-12s minigene
To identify potential splicing regulator binding sites in MDM2 transcripts (Fig. 1A), we scanned the coding sequence of the MDM2 full-length mRNA with ESEfinder (19). Two of the most significant hits were for a pair of SRSF2-binding sites in exon 11 of MDM2. We then identified changes in SRSF2-binding sites that lowered matrix-binding site scores (Fig. 1B) and subsequently induced these mutations. We transfected the wild-type and mutant MDM2 3-11-12s minigenes into MCF7 cells, then treated under normal, UVC, or cisplatinum conditions. RT-PCR analysis revealed that the wild-type MDM2 3-11-12s minigene maintained full-length 3.11.12 splicing under normal conditions, and UVC or cisplatinum treatment induced the skipped product 3.12. However, mutation of either SRSF2 site (G165T or G213T) in exon 11 resulted in the increased expression of the exon-excluded product in the absence of damage as compared with the wild-type minigene (Fig. 1C and D; Supplementary Fig. S1A and S1B). In addition, mutation of both sites together (G165T, G213T) had an additive effect on exon exclusion under normal conditions, indicating that both sites function in the recognition of exon 11 by the spliceosome.
SRSF2 is relocalized in the nucleus and has decreased binding to MDM2 exon 11 in response to UVC treatment
Alternative splicing of MDM2-ALT1 is induced under conditions of genotoxic stress and is coincident with an increase in SRSF1 protein expression (3). We therefore hypothesized that there may be a decrease in the expression of SRSF2 under the same conditions. Surprisingly, we observed that expression of SRSF2 in MCF7 and HeLa cells increased after UVC treatment (Fig. 2A; Supplementary Fig. S2). We then hypothesized that sequestration of SRSF2 in nuclear speckles was a potential mechanism of preventing its availability to regulate MDM2 alternative splicing. Thus, we treated MCF7 and HeLa cells with UVC and performed immunofluorescence against SRSF2. Beginning at approximately 4 hours, SRSF2 nuclear speckles became larger (Fig. 2C; Supplementary Fig. S3A and S3B), which correlates with the timing of MDM2-ALT1 induction (3). Importantly, we found that the average size of nuclear speckle foci was significantly larger after 12 hours of UVC exposure (Fig. 2B; Supplementary Fig. S3E). We also performed costaining of nuclear speckle marker SRRM2 and demonstrated that there was colocalization of SRSF2 and SRRM2 at 0 and 12 hours after UVC treatment (Supplementary Fig. S3C and S3D; Supplementary Movies SRSF2 0h UVC and SRSF2 12h UVC). In addition, we report that there was a significant increase in sphericity, or roundness, of SRSF2 foci at 12 hours as compared with 0 hours of UVC exposure (Supplementary Fig. S3F). These data suggest that SRSF2 relocalization is coincident with MDM2-ALT1 expression and thus SRSF2 may be playing a direct role in facilitating MDM2 splice site selection in the absence of damage treatment.
To determine whether relocalization of SRSF2 is concurrent with decreased binding to exon 11 of the endogenous MDM2 pre-mRNA after damage treatment, we performed RNA immunoprecipitation with and without UVC treatment. SRSF2 bound exon 11 under normal conditions; however, this binding was significantly attenuated in conditions of UVC stress (Fig. 2D and E). Furthermore, SRSF2 did not bind to the negative isotype control (mIgG), and we did not observe any amplification of DNA in our control reactions that were not reverse transcribed. We also tested a positive control, CCNL1 exon 4 (20). Like MDM2 exon 11, SRSF2 bound less to CCNL1 exon 4 under UVC stress as compared with normal conditions (Supplementary Fig. S4). These results demonstrate that the relocalization of SRSF2 under damage conditions may change availability of binding and splicing behavior of several transcripts in addition to MDM2.
To further demonstrate that ESE mutation leads to abrogated binding of SRSF2 we performed in vitro RNA pull down assays using both wild-type and mutant sequences of each binding site. Our results confirmed that SRSF2 does in fact bind to both conserved predicted binding sites and mutation of either position significantly abrogated its binding in vitro (Fig. 2F and G) and in vivo.
SRSF2 is a positive regulator of MDM2 alternative splicing
To assess the role of SRSF2 as a positive regulator of MDM2 alternative splicing, we performed overexpression and knockdown experiments. We cotransfected MCF7 cells with the MDM2 3-11-12s wild-type minigene and either LacZ or T7-SRSF2, followed by normal or UVC treatment (Fig. 3A). We observed that transfection of SRSF2 abolished damage-responsive alternative splicing of minigene transcripts under UVC conditions as compared with the negative control (Fig. 3B). We also demonstrated that SRSF2 overexpression is not confined to nuclear speckles under UVC treatment and therefore is not subject to the same regulation as endogenous SRSF2 (Supplementary Fig. S5).
Conversely, we treated MCF7 cells with siRNA against SRSF2 (Fig. 3C). We then performed a nested RT-PCR to identify transcripts from the endogenous MDM2 gene. We found that MDM2-ALT1 (3.12) was significantly induced in the absence of any genotoxic stress as compared with no treatment and control siRNA (Fig. 3D). These data indicate that SRSF2 expression is required to facilitate inclusion of all MDM2 internal exons (exons 4 through 11), not only the penultimate exon 11 as studied in our minigene system.
SSOs against SRSF2 sites in exon 11 induce expression of MDM2-ALT1
One of the most common therapies to induce changes in splicing activity in disease is the use of SSOs. SSOs are single-stranded RNA molecules that bind to complementary SRE (splicing regulatory element) targets to occlude binding of a specific RNA-binding factor, thereby impacting the levels of splicing in a target transcript. We hypothesized that SSOs targeting SRSF2-binding sites would occlude SRSF2 binding and could therefore be used to induce skipping of the internal exons of MDM2. To test this hypothesis, we designed SSOs against each of our identified binding sites (Fig. 4A). We cotransfected the wild-type MDM2 3-11-12s minigene along with SSOs against SRSF2 sites in MCF7 cells and observed that SSOs against either SRSF2 site were effective at inducing expression of 3.12 under normal conditions compared with nonspecific SSO (NS-SSO; Fig. 4B). Overall, SSO2 and SSO3 were more potent in inducing exon exclusion as compared with SSO1 (Fig. 4C). These data are consistent with mutation of the SRSF2-binding sites in the MDM2 minigene described above.
Decreasing the levels of SRSF2 caused skipping of multiple internal exons of the endogenous MDM2 transcript, generating MDM2-ALT1. We therefore hypothesized that SSOs targeting the SRSF2-binding sites in exon 11 could likewise induce MDM2-ALT1 expression. We transfected SSOs into MCF7 or SMS-CTR cells. Consistent with our results with MDM2 3-11-12s minigene, we were able to observe a significant increase in expression of MDM2-ALT1 (Fig. 4D and E).
We assessed the functional impact of SSO treatment on the p53 pathway by examining transcriptional target levels, as well as cell-cycle changes. We found that SSOs targeting the second SRSF2 site (SSO2, SSO3) significantly reduced the expression of GADD45A and CDKN1A in MCF7 cells (Fig. 4F). MCF7 cells are ARF null and therefore lack an intact p53 pathway (21), which may explain the lack of induction of p53 targets in response to MDM2-ALT1 expression.
We subsequently performed SSO treatment in rhabdomyosarcoma cells, Rh18, which are p53 wild-type and express high levels of MDM2. We observed similar levels of MDM2-ALT1 induction in response to SSO treatment and also a significant increase in p53 protein expression (Fig. 4G). In addition, we observed an increase in cells undergoing early apoptosis for all MDM2 exon 11 SRSF2 SSOs (Fig. 4H). Consistent with these findings, we also found a significant increase in proapoptotic gene PUMA, as well as p53 target WIP1 (Fig. 4I). Altogether, these findings underscore the therapeutic potential of activating the p53 pathway in cancer through MDM2 alternative splicing in cells with overexpressed MDM2 that maintain wild-type p53.
SRSF2 regulation of MDM2 splicing is conserved from mouse to human
Like the human MDM2 gene, mouse Mdm2 undergoes alternative splicing under genotoxic stress. The predominant splice variant, Mdm2-MS2, however, comprises exons 3, 4, and 12 (Fig. 5A; ref. 3). To determine the conserved regulation of MDM2 splicing by SRSF2 regulation, we generated a mouse minigene that contains the Mdm2 exon 3 as the first exon joined to exon 11 and exon 12. We next examined the sequence of mouse exon 11 to identify mutations that lower the ESE matrix scores for predicted SRSF2-binding sites in the mouse gene. Mutations similar to those we made in the human minigene lowered the predicted binding scores of the ESE in Mdm2 exon 11 (Fig. 5B). We then induced the G165T or G213A mutations, or both, and tested their ability to support Mdm2 splicing regulation. Mutations of both the G165T and G213A residues resulted in increased expression of exon 11-excluded product in the absence of genotoxic stress (Fig. 5C and E). As with the MDM2 3-11-12s G213T mutation, the Mdm2 3-11-12s G213A mutation was more potent than the G165T mutation.
Because the mouse splice variant, Mdm2-MS2, is comprised of exon 4 spliced directly to exon 12, we wondered whether the sequences in Mdm2 exon 4 may be important to achieve regulation in response to damage treatment in the mouse gene. We engineered a second mouse minigene containing exons 4, 11, and 12 and induced these same mutations. We observed that both G165T and G213A mutations together resulted in an increase of exon skipping under normal conditions (Fig. 5D and F), and were induced by UVC treatment. The effects of the G165T and G213A mutations were additive in both the 3-11-12s and 4-11-12s minigene, suggesting that these SRSF2 sites regulate Mdm2 splicing independently.
CRISPR-Cas9–engineered mutant cell lines demonstrate endogenous regulation of Mdm2 by SRSF2 sites in exon 11
To pinpoint the regulation of SRSF2 on Mdm2 endogenous transcripts we used CRISPR-Cas9 to induce SRSF2 site mutations. We designed Cas9 guide RNAs to exon 11 of Mdm2 as well as single-stranded oligonucleotide donor (ssODN) repair templates that included the mutation at our sites of interest (Fig. 6A). We recovered NIH 3T3 cell lines with the single G165T mutation, but were unable to generate cell lines with the G213A mutation. We report that cells with the G165T mutation demonstrated a significantly higher amount of Mdm2-MS2 under normal conditions as compared with control cells (Fig. 6B).
To test the effect of Mdm2-MS2 expression on cell proliferation, we monitored CRISPR-engineered cell lines continuously for growth. At the end of 72 hours, both of our G165T clones showed significantly higher cell confluence (CRISPR1 and CRISPR2) in comparison with the wild-type CRISPR control (CRISPR CTRL, Fig. 6C). It is important to note that NIH 3T3, like MCF7 cells, although wild-type for p53, are mutant for ARF. In the absence of an intact p53 pathway, MDM2 splice variants have p53 independent functions that support increased cell proliferation and transformation. Indeed, it has been demonstrated by others that MDM2 splice variants are capable of transforming NIH 3T3 cells (22).
SRSF1 and SRSF2 are antagonistic for the control of MDM2 splicing
Given that the identified SRSF2 sites flank the previously identified SRSF1-binding site, we wanted to test the dependency of these sites in counterbalancing the activity of each other. We hypothesized that the positive action of the SRSF2 binding is required to overcome the negative splicing function of the SRSF1 binding. We induced the previously published SRSF1 mutation together with one or both of the SRSF2 mutations (Fig. 7A). We observed that both the SRSF2-G165T and the SRSF2-G213T mutation overcame the mutation of the SRSF1 sites, and restored the damage induction of exon 11. These data suggest that the regulation of MDM2 alternative splicing by SRSF2 is necessary to counteract the negative effects of SRSF1. However, when both SRSF2 sites were mutated in the context of the SRSF1 mutant we again observed a significant increase in the level of exon skipping under normal conditions (Fig. 7B and C). Because splicing control of MDM2 is maintained in the absence of both the SRSF1 and SRSF2 elements, identification of additional elements will allow improvement of splice-altering therapies that could be used to modulate the p53 pathway in cancer.
We have identified SRSF2 as a positive splicing factor that promotes the recognition of exon 11 of MDM2 and demonstrated that SRSF2-binding sites are conserved between mouse and human MDM2 exon 11. Furthermore, these sites are sufficient to promote full-length splicing endogenously as splicing is compromised using either the SSOs or CRISPR-Cas9–generated mutations. Exon 11 of MDM2 is well-conserved between mouse and human. Both are 78 base pairs and share approximately 82% nucleotide identity. While both SRSF2 binding sites are conserved according to in silico prediction (23), the first site has a mismatch at 3/8 positions and the second is 100% conserved. In both mouse and human minigene mutations, as well as in SSO treatment, the second SRSF2 site consistently induced more exon exclusion compared with the first and may explain the pressure for it to remain conserved over evolutionary time.
The regulation of SRSF2 is known to be controlled through alternative splicing of its own transcript, as well as posttranslational acetylation and phosphorylation by SRPK1 and SRPK2 (24). Under both normal and UVC damage conditions, SRSF2 is localized to nuclear speckles, which are known to be structures where nuclear processing factors are localized, as well as sites of active mRNA transcription (25, 26). In response to UVC treatment, we observed that the expression levels of SRSF2 increased and importantly, that the speckles had a larger foci diameter. In addition, we saw that binding of SRSF2 was decreased in vivo in response to UVC treatment. We infer that the loss of colocalization of SRSF2 with MDM2 transcripts under conditions of genotoxic stress facilitates MDM2 alternative splicing.
MDM2 is overexpressed in many types of cancer including osteosarcoma, esophageal cancer, and dedifferentiated liposarcomas (1, 27, 28). These cancers are invariably p53 wild-type and would benefit from persistent alternative splicing of MDM2 to downregulate MDM2 expression, reactivate p53, and sensitize these tumors to current therapies. Whereas other drugs including nutlin-3a and spiro-oxindole analogues MI-63 and MI-219 have shown potential by disrupting the MDM2-p53 interaction as anticancer strategies (29–31), these particular compounds have not been successful in the clinic likely due to their failure to also inhibit MDMX. Our lab previously demonstrated that MDM2-ALT1 can bind both full-length MDM2 and MDMX and sequester these proteins in the cytoplasm, thereby acting as a dominant negative to the function of full-length MDM2 or MDMX (3, 32). Promoting the alternative splicing of MDM2 through the use of SSOs would allow for control of both full-length MDM2 and MDMX activity, thereby elevating the levels of p53 in a cancer cell.
While MDM2 overexpression in cancer is well-documented in the literature, paradoxically MDM2 alternative splicing has been observed with many types of cancer as well, including bladder (22), colon (33), breast (34), and soft tissue sarcomas such as rhabdomyosarcoma (35). We previously reported that MDM2-ALT1 expression correlated with high-grade disease in rhabdomyosarcoma and is the most common molecular genetic perturbation in both alveolar and embryonal rhabdomyosarcoma. Recent data suggests that the mechanisms of MDM2-ALT1-mediated oncogenesis are largely p53-independent as expression of MDM2-ALT1 in a p53-deficient background accelerated tumorigenesis and shifted the observed tumor spectrum in vivo (36). Therefore, while it would be beneficial to induce alternative splicing of MDM2 in cancers where it is overexpressed, in cancers with p53 gain-of-function mutations it may be useful to restore full-length MDM2 to degrade mutant p53. As we have now characterized both positive and negative regulators of MDM2 alternative splicing, SSO therapy could be used to modulate the levels of p53 through the manipulation of MDM2 alternative splicing.
Disclosure of Potential Conflicts of Interest
D.S. Chandler is a consultant/advisory board member for Vironexis. No potential conflicts of interest were disclosed by the other authors.
Conception and design: D.F. Comiskey Jr, M. Montes, S. Khurshid, R.K. Singh, D.S. Chandler
Development of methodology: D.F. Comiskey Jr, M. Montes, S. Khurshid, R.K. Singh, D.S. Chandler
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.F. Comiskey Jr, M. Montes, S. Khurshid, R.K. Singh, D.S. Chandler
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.F. Comiskey Jr, M. Montes, S. Khurshid, R.K. Singh, D.S. Chandler
Writing, review, and/or revision of the manuscript: D.F. Comiskey Jr, M. Montes, S. Khurshid, R.K. Singh, D.S. Chandler
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.F. Comiskey Jr, M. Montes, S. Khurshid, D.S. Chandler
Study supervision: D.S. Chandler
This work was funded by NCI grant R01 CA133571 (to D.S. Chandler) and a Pelotonia fellowship (to D.F. Comiskey). We would like to thank Aixa S. Tapia-Santos and Jordan T. Gladman for their technical contribution. We would also like to thank the members of the Chandler lab for critical review of the manuscript. Images presented in this report were generated using the instruments and services at the Campus Microscopy and Imaging Facility, The Ohio State University (Columbus, OH). This facility is supported in part by NCI grant P30 CA016058.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.