Genotoxic Stress Induces Coordinately Regulated Alternative Splicing of the p53 Modulators MDM2 and MDM4

The tumor suppressor protein p53 is a transcription factor that induces G₁ arrest of the cell cycle and/or apoptosis. The murine double-minute protein MDM2 and its homologue MDM4 (also known as MDMX) are critical regulators of p53. Altered transcripts of the human homologue of mdm2, MDM2, have been identified in human tumors, such as invasive carcinoma of the breast, lung carcinoma, and liposarcoma. MDM2 alternate forms act to negatively regulate the normal MDM2 gene product, thus activating p53. Although many reports have documented a plethora of tumor types characterized by MDM2 alternative transcripts, few have investigated the signals that might initiate alternative splicing. We have identified a novel role of these alternative MDM2 transcripts in the normal surveillance mechanism of the cell and in DNA damage response. We report that alternate forms of MDM2 are detected after UV irradiation. Furthermore, we show that mouse cells treated with UV are also characterized by alternative transcripts of mdm2, suggesting that this is an important and evolutionarily conserved mechanism for regulating the expression of MDM2/mdm2. An additional p53 regulator and mdm2 family member, MDM4, is likewise alternatively spliced following UV irradiation. By activating alternative splicing of both MDM2 and MDM4, yet another layer of p53 regulation is initiated by the cells in response to damage. A stepwise model for malignant conversion by which alternate forms of MDM2 and MDM4 place selective pressure on the cells to acquire additional alterations in the p53 pathway is herein proposed. (Cancer Res 2006; 66(19): 9502-8)

p53 is a transcription factor able to induce G₁ arrest of the cell cycle and/or apoptosis by transactivating numerous downstream genes. p53 becomes stable and active after genotoxic stress through modifications, such as phosphorylation and acetylation. Several proteins are reported to affect p53 activity, but the murine double-minute protein MDM2 and a closely related family member, MDM4, are two of the most critical regulators of p53. mdm2 was discovered by virtue of its amplification in the spontaneously transformed cell line 3T3DM (1). MDM2 directly binds to p53 and blocks its transcriptional activity (2–4) as well as acts as a E3 ubiquitin ligase (5), promoting p53 localization to the cytoplasm and subsequent proteasomal degradation (6, 7). The importance of the role of mdm2 in the regulation of p53 activity and stability is underscored by knockout mouse experiments, in which the early lethality of the mdm2-null embryos is rescued in mice that are also p53 null (8, 9). Alternatively spliced isoforms of MDM2 that lack the p53-binding domain but retain the RING domain have been detected in human and mouse tumors and are capable of repressing normal mdm2 down-regulation of p53 and to inhibit proliferation. Specifically, splice variants of the human mdm2 homologue, MDM2 (or HDM2), have been identified in pediatric rhabdomyosarcoma cell lines and tumors at a frequency of 75% and 82%, respectively (10). Alternate MDM2 transcripts have also been identified in liposarcomas, gliomas, and lung and breast carcinomas (11–13).

Like MDM2, MDM4 can also bind to p53 and inhibit its transcriptional function. MDM4-null mice die early in embryogenesis and can also be rescued by concomitant deletion of p53 (14). Most recently, MDM4 alternative transcripts have likewise been reported in soft tissue sarcomas, allowing for the possibility that alternative forms of MDM4 negatively regulated the full-length MDM4 protein in a similar fashion (15). The alternatively spliced forms of MDM2 lacking p53-binding domain and ARF domain can interact with MDM2 and inhibit its activity, which in turn up-regulates p53 activity (11). A similar aberrantly alternatively spliced form of MDM4, HDMX211 (exons 2 and 11), also lacks the p53-binding region and has been shown to stabilize the p53 protein by counteracting its degradation by MDM2 (16).

Because negative regulation of either MDM2 or MDM4 would activate the p53 tumor suppressor pathway, the role of these transcripts in tumorigenesis seems contradictory. We therefore proposed that the alternative MDM2 and MDM4 transcripts play a role in the normal surveillance mechanism of the cell. These cumulative studies provide evidence to suggest that the production of MDM2 and MDM4 alternative transcripts is indeed triggered by genotoxic stress and may help activate the p53 protective mechanism.

Cell culture, plasmids, radiation, and drug treatment. MCF7 (human breast carcinoma, p53 wild-type), H1299 (human lung carcinoma, p53 null), U2OS (human osteosarcoma cell line, p53 wild-type), SJSA-1 (human osteosarcoma, MDM2 overexpressing), and NIH3T3 (mouse fibroblast) cells were obtained from the American Type Culture Collection (Manassas, VA). Wild-type, ARF^−/−, and p53^−/− mouse embryonic fibroblasts (MEF) were generated by crossing mice of the appropriate genotypes and collecting embryos at 13.5 d.p.c. Experiments using MEFs were carried out at early passages (≤P5). Cell lines were maintained in DMEM supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), 2 mmol/L l-glutamine, and 0.1 mg/mL penicillin (100 IU/mL)/streptomycin (100 μg/mL) at 37°C in a humidified atmosphere of 5% CO₂. Cell culture medium and additives were obtained from Invitrogen (Carlsbad, CA) if not otherwise stated. For damage treatment, cells were plated at 50% to 60% confluency, grown at 37°C overnight, and treated with the appropriate amount of UV cross-linker (Stratalinker 1800, Stratagene, La Jolla, CA), ionizing radiation (IR; IBL 437C H-type, CIS Bio International, Gif-sur-Yvette, France), or cis-platinum. Cisplatin was obtained from Sigma (St. Louis, MO; P4394), and stock solutions were prepared at 0.5 mg/mL in PBS. The cells were allowed to continue to grow at 37°C and harvested for total RNA and protein at specific time points after treatment.

Trypan blue exclusion assay and determination of equitoxicity of radiation treatment. MCF7 cells were seeded 1 × 10⁶ in a 10-cm tissue culture dish, grown at 37°C overnight, and treated with 6, 12, or 25 Gy of ionization or 10, 30, or 50 J/m² of UV radiation. Cell viability was counted 24 hours after radiation treatment by trypan blue exclusion. Percentage viability was calculated and normalized to untreated cells in triplicate, and SD was calculated.

Nested reverse transcription-PCR and Western blot analysis. Total RNA was isolated using Qiagen RNeasy (Valencia, CA) and subjected to reverse transcription using First-Strand cDNA synthesis kit (Amersham, Piscataway, NJ). The human MDM2 transcripts were then amplified from the resultant cDNA using a nested set of primers as reported previously (11). Other cDNAs were amplified using a similar nested PCR strategy with gene-specific primer as follows: For MDM2, the first PCR primer pair was CTTCTCGTCGCTCGAGCTCTGGAG (MDE1F-5910) and TCTGCACACTGGGCAGGGCTTGTTT (MDE12R-1425) and the nested primer pair was CCTCCCAGGCCAATGTGCAATAC (MIF2-3 MDM2) and CATTTGGATTGGCTGTCTGC (MIR1). For human p53, the first PCR primer pair was TGCCTTCCGGGTCACTGCC (p53X2exS) and CTGTCAGTGGGGAACAAGAAGTG (Hp533′UTRas) and the nested pair was CATGGAGGAGCCGCAGTCAG (hp53X2inS) and GTCAGGCCCTTCTGTCTTGAAC (Hp53 X11as). For human O⁶-methylguanine methyltransferase (O⁶-MGMT), the first PCR primer pair was GGACAAGGCTTGTGAAATGAAAC (MGMTExSn) and CGTCAAACATCCATCCTACTGC (MGMTUTRas) and the nested primer pair was CTCCTGGGCAAGGGGACGTCTG (MGMTInSn) and GTAGCTCCCGCTCCCTTGAGCCAG (MGMTInAsn). For MDMX, the first PCR primer pair was CATTTTCCACCTCTGCTCAGTGTTC (HDMX ExSn) and CAAATAGGGCATGAAGCCCCAGC (HDMX ExAsn) and the nested primer pair was CAATCAGGTACGACCAAAACTGCC (HDMX InSn) and GAGATGGTCTCTTGGCTTCAGAAC (HDMX InAsn). PCR products were resolved on a 1.2% agarose gel, TA cloned (Invitrogen), and sequenced for positive identification using vector-specific TAATACGACTCACTATAGGG (T7) or ATTTAGGTGACACTATAG (SP6) primers. For Western blot analysis, cells were lysed using laemmli buffer. Protein lysates were separated by 10% SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane (Hybond-P, Amersham), and Western blot analysis was done using the p53 using DO-1 (p53-specific antibody; Santa Cruz Biotechnology, Santa Cruz, CA) and actin (a-5541) antibody (Sigma).

Inhibition of ataxia-telangiectasia mutated/ataxia-telangiectasia mutated and Rad3-related kinases. MCF7 cells were treated with 0, 0.2, and 5 mmol/L caffeine (Sigma) from a freshly made stock (50 mmol in DMEM complete medium) for 2 hours before UV treatment. Cells were maintained in corresponding amount of caffeine after treatment and harvested 6 hours after UV treatment. Cells were lysed using NP40 lysis buffer [10 mmol/L Tris (pH 7.5), 250 mmol/L NaCl, 5 mmol/L EDTA, 50 mmol/L NaF, 1 mmol/L Na₃VO₄, 1 mmol/L phenylmethylsulfonyl fluoride, 1% NP40, 1× protease inhibitor], and 30 μg of total protein were separated on 10% SDS-PAGE for each samples. Proteins were transferred to a PVDF membrane (Hybond-P), and Western blot analysis was done with phosphorylated p53 Ser¹⁵ antibody [ataxia-telangiectasia mutated (ATM) and ATM and Rad3-related (ATR) specific phosphorylation site; Cell Signaling Technology, Danvers, MA], DO-1 (p53 specific antibody), and actin (a-5541) antibody.

Cloning, transfection, and immunohistochemistry. The UV-induced form of MDM2, MDM2^alt1, was cloned into a mammalian expression vector, pCCALL2 (17), driven by the cytomegalovirus (CMV) promoter and the chick β-actin enhancer with an NH₂-terminal myc tag. Cre-induced recombination in BNN132 Escherichia coli cells removed the βgeo cassette from the pCALL2 to obtain pCALL2-myc-MDM2^alt1. NIH3T3 cells were plated at 65% confluency in a 6-cm dish, transfected with 2 μg pCCALL2-myc-MDM2^alt1 plasmid using Fugene 6 (Roche, Nutley, NJ), fixed 24 hours after transfection, and stained for immunofluorescence as described previously (11). The transfected human protein was visualized using a myc antibody (red), whereas the endogenous mouse MDM2 was visualized using the SMP14 antibody (green). Nuclei were visualized using 4′,6-diamidino-2-phenylindole (DAPI) stain (Vector Laboratories, Burlingame, CA). Photographs were taken at ×400 magnification using a Leica DM4000B (Bannockburn, IL) microscope equipped with a digital camera, SPOT RT KE, and SPOT advance system software (Diagnostic Instruments, Inc, Sterling Heights, MI).

Induction of MDM2 alternative splicing in human cell lines. To test the hypothesis that MDM2 alternative transcripts play a role in cellular surveillance, we treated human cell lines with various types of genotoxic stress (approximate equitoxic doses of UV and IR) and assayed the cells for their ability to produce alternative transcripts of MDM2. For the human breast carcinoma cell line MCF7 cells, 10 Gy IR is approximately equitoxic to 10 J/m² UV irradiation (18). To determine the approximate biological equivalent doses for the remaining UV and IR irradiation treatments, we did trypan blue exclusion assays 24 hours following treatment of MCF7 cells. We found that 30 J/m² UV was approximately equivalent to 12 Gy IR (55.25 ± 9.2% and 50.61 ± 14.62% viability, respectively) and 50 J/m² was approximately equivalent to 25 Gy IR (40.85 ± 13.41% and 38.22 ± 4.33% viability, respectively). We therefore treated the MCF7 cells with the appropriate doses of UV (0, 10, 30, and 50 J/m²) or IR (0, 6, 12, and 25 Gy) and assayed the ability of the cells to support MDM2 alternative splicing. Because MDM2 is normally expressed in low levels in the cell, we used a sensitive, yet nonquantitative, nested reverse transcription-PCR (RT-PCR) strategy to visualize the full-length and alternate transcripts. Treatment of the human breast carcinoma cell line MCF7, which is p53 wild-type, induced alternative splicing of the MDM2 gene with either 30 or 50 J/m² as early as 6 hours after treatment of the cells (Fig. 1A). Expression of the spliced variant peaks 36 hours after treatment and subsequently subsides in the next 24 hours with no spliced form visible at 72 or 168 hours (1 week) after treatment (Fig. 1D; data not shown). This same splicing pattern for MDM2 was also observed in response to UV treatment in the human wild-type fibroblast cell line WI38 (data not shown). The predominant alternate spliced form of MDM2 was cloned and sequenced and identified as the MDM2^alt1 (also known as HDM2-B) that was previously identified in numerous cancer types. Other known alternative transcripts of MDM2, including MDM2-A, MDM2-C, and MDM2-KB, were also induced in response to UV to a much lesser extent (faint bands visible in Fig. 1A,, right). Strikingly, no alternative splicing of MDM2 was obtained in MCF7 or H1299 cells treated with approximately equitoxic doses of IR (Fig. 1A; data not shown).

Smaller PCR products were frequently seen in all cell lines analyzed (data not shown). These products were not damage inducible or reproducible from experiment to experiment. Nonetheless, these products were cloned and sequenced and found to correspond to the “aberrant” spliced forms that have been reported (10, 13). Sequence analysis of the resultant MDM2 cDNA and comparison with the MDM2 genomic DNA do not reveal donor and acceptor splice sites or even an appropriately positioned polypyrimidine tracts and branchpoints at which the catalysis of the RNA splicing reaction can be initiated. It seems unlikely, then, that these cDNAs arise from alternatively spliced pre-mRNA. Given that the aberrant PCR products have repetitive sequences at their junctions in place of bona fide splice sites and that their presence is not consistent from experiment to experiment, we conclude that these products are an unfortunate artifact of the highly sensitive nested PCR strategy. We further deduce that their presence is overrepresented as they are disproportionately amplified due to their small size. There, however, remains a remote possibility that these aberrant products are the result of a novel cellular process, but because these products were not damage inducible, they were not investigated further.

Damage pathways and alternative MDM2 splicing. Because MDM2 is regulated at the transcriptional level by p53 and transcription and splicing are intricately linked, we wanted to determine the effect of p53 on the regulation of MDM2 splicing. We used a p53^−/− cell line (human lung carcinoma cell line, H1299) and tested for a splicing alteration in response to damage. As shown in Fig. 1A, the lack of p53 did not affect MDM2 alternative splicing in response to damage with alternate transcripts being expressed 12 and 24 hours after treatment with either 30 or 50 J/m² of UV treatment. Furthermore, to determine if alternative splicing of the MDM2 transcript is a mechanism for the cells to regulate the absolute levels of MDM2 in response to its overexpression, we looked for alternatively spliced products in the human osteosarcoma cell line SJSA-1, in which MDM2 is amplified and overexpressed (19). There was no expression of MDM2 alternative transcripts in this cell line (Fig. 1C).

The cellular response to damaged DNA induced by UV (pyrimidine dimers) is distinct from the cellular response to γ irradiation (double-strand breaks). As mentioned previously, MCF7 and H1299 cells that were treated with increasing amounts of IR did not show an induction of MDM2 alternative transcripts. To confirm the placement of MDM2 splicing in a specific damage pathway, cells were also treated with another damaging agent, cis-platinum. Cis-platinum, like UV, causes helix-distorting lesions by DNA cross-linking. Three cell lines, MCF7, H1299, and U2OS (human osteosarcoma) cells, were treated with increasing amount of cisplatin to induce damage. MDM2 alternative splicing is induced by cis-platinum in all three cell lines tested (Fig. 1C).

Generally, the ATM pathway is induced in response to double-strand breaks in DNA as a result of IR treatment and the ATR pathway is induced in response to helix-distorting lesions, such as those incurred by UV or cisplatin treatment (20). Because the kinases ATM and ATR are integral signaling effectors for the induction of p53 in response to DNA damage, we wanted to test their ability to affect splicing of MDM2. To do this, we treated MCF7 cells with caffeine to block both ATM and ATR kinase activity and treated the cells with 50 J/m² to induce alternative splicing of MDM2. As shown in Fig. 2, damage-responsive alternative splicing of MDM2 persists even in the absence of ATM and ATR signaling (5 mmol/L caffeine).

Induction of mdm2 alternative splicing in mouse cell lines. Alternatively spliced forms of mdm2 have likewise been reported in Eμ-myc-induced lymphomas (21). Furthermore, expression of a mouse alternative spliced form analogous to MDM2^alt1 can lead to the formation of myeloid sarcomas in transgenic mouse studies (22). We therefore wanted to know if the mouse mdm2 gene undergoes alternative splicing in response to UV damage in the same way that human MDM2 does. We treated several mouse cell lines, including wild-type MEFs, p53^−/− MEFs, and NIH3T3 fibroblasts. Shown in Fig. 3A are representative results obtained from MEFs that are null for p53. Two predominant mdm2 spliced forms are obtained in response to damage. These alternative spliced forms of mdm2 were cloned, sequenced, and designated MS1 and MS2 (Fig. 3C). Once again, the alternative splicing of mdm2 is independent of p53 due to the fact that p53-null cells retain their ability to express the alternate forms. We additionally wanted to test the ability of the known MDM2 regulator, ARF, to affect alternative splicing of mdm2. To do this, we treated ARF^−/− MEFs with UV and used nested PCR to visualize the mdm2 RNA species present. As shown in Fig. 3B, the absence of ARF (ARF^−/− MEFs) does not affect splicing of mdm2, as the alternative spliced forms of mdm2 prevail in these cells.

The alternatively spliced forms of mdm2, MS1 and MS2, are analogous to the human MDM2 transcripts, as they lack the p53-binding domain but retain the COOH-terminal RING domain. To determine whether the mouse MDM2 can be relocalized to the cytoplasm as a conserved mechanism of protein regulation, we did transfection experiments in NIH3T3 cells. After MDM2^alt1 transfection, protein localization was visualized by immunofluorescence (Fig. 3D; Supplementary Fig. S1). In nontransfected cells, MDM2 was localized primarily in the nucleus. The transfected cells, however, show MDM2 relocalized in the cytoplasm, colocalizing with the myc-MDM2^alt1. This confirms that, in the mouse, MDM2 is subject to regulation by binding to and relocalization by alternative proteins that lack the p53-binding domain and nuclear localization signal. Strikingly, the levels of endogenous MDM2 were distinctly elevated in cells transfected with MDM2^alt1.

Induction of MDM4 alternative splicing. In human cell lines, at least three different isoforms of MDM4 alternative transcripts have been identified (23). As with MDM2, alternate spliced forms of MDM4 were detected in soft tissue sarcomas (15). Because MDM4 is structurally and functionally similar to MDM2, we wanted to determine if alternative splicing likewise plays a role in its regulation in response to damage. We found that two predominant alternative transcripts of MDM4 are indeed produced at varying doses and times after UV irradiation (Fig. 4A). Sequencing of these transcripts showed that XALT2 is similar to spliced forms of MDM2 in that it also lacks the p53-binding domain and retains the COOH-terminal RING domain (Fig. 4D). The MDM4 splice variant 1 (XALT1) is completely unique in that only the p53-binding domain is maintained in the resultant protein (Fig. 4D). Other spliced forms of MDM4 were also present at very low levels in damage treated cells. These forms included HDMX-A (24) and HSMX-S (25), some of which have been identified in human cancers (15).

Because the cells undergo dramatic changes and turn on numerous signaling pathways in response to damage, it is possible that splicing is globally disrupted in a cell of that state. To test this, we analyzed the splicing of two known DNA damage response genes, p53 and O⁶-MGMT, after cis-platinum treatment of cells. Shown in Fig. 4B and C are results obtained using three human cell lines. Damage-dependent alternatively spliced forms were not seen for p53 or the DNA repair gene O⁶-MGMT. No transcripts were seen for p53 in H1299 cells because these cells are p53 null. The one smaller p53 band seen in U2OS treated with 25 μmol/L cis-platinum as well as two other smaller bands (data not shown) were sequenced and determined to not be a p53 spliced variants but rather an artifact of the reverse transcription reaction with repetitive sequences at the junction points as seen with MDM2. It is worth noting that, although the primer sets used should pick up the new variant of p53 recently reported (26), we do not see this variant in any of the cell lines analyzed.

In this study, we have shown that alternative splicing of two regulators of p53, MDM2 and MDM4, is initiated in a coordinately regulated fashion in response to cellular stress. The MDM2 pre-mRNA shows a dramatic alteration in splicing pattern that yields predominantly the MDM2^alt1 (MDM2-b) spliced isoform along with other similar spliced forms that also lack the p53-binding domain but retain the COOH-terminal RING domain. Coincidentally, MDM2^alt1 is the most commonly identified alternative MDM2 transcript in astrocytic neoplasms, ovarian cancers, and breast cancers. Significantly, this alternative form lacks not only most of the p53-binding domain but also both the nuclear localization and export signals as well as the ARF-binding domain (Fig. 1B). This form has previously been shown not only to be localized to the cytoplasm itself but also to direct the mislocalization of the full-length MDM2 protein to the cytoplasm presumably by direct protein interaction (11) and, as a result, increase protein levels of p53 (Fig. 1D), thereby inducing the p53 pathway. In this way, induction of alternative splicing of MDM2 could help to induce the p53 damage response.

Our data indicate that alternative splicing of MDM2 is responsive to helix-distorting lesions of DNA, such as occurs from UV or cisplatin treatment but not to double-strand breaks as from IR irradiation. This dichotomous response to different damaging agents mirrors what is reported for the differential specific phosphorylation of p53 that occurs in response to UV but not γ irradiation (27). Factors that direct alternative splicing are similarly controlled via their phosphorylation status (28) and may therefore be regulated by some of the same signaling pathways as p53. Importantly, factors that regulate the splicing of MDM2 in response to genotoxic stress may be downstream effectors of the ATR and nucleotide excision repair pathway.

To begin to understand the molecular mechanisms used to achieve regulated splicing of the MDM2 transcript, we used genetically deficient cells or pharmacologics to delete or inhibit and thus interrogate specific damage response pathways. Because p53-deficient cells are capable of triggering alternative splicing of MDM2 in response to damage, we concluded that the pathway is controlled independently of p53. To determine if the regulated splicing of MDM2 is part of a feedback loop activated by increased levels of MDM2 expression to control the absolute levels of MDM2 expression, we queried a cell line known to have increased levels of MDM2. MDM2 overexpression alone does not initiate alternative splicing of the MDM2 transcript because the cell line SJSA-1, in which MDM2 is amplified and overexpressed, does not constitutively express alternatively spliced forms of MDM2. Additionally, we can conclude that control of MDM2 splicing is independent of ATM and ATR kinase activity because blocking these pathways with caffeine had no effect on induction of alternative splicing.

Our data indicate that mouse cells are also capable of initiating alternative splicing of mdm2 in response to specific types of genotoxic stress, underscoring the importance of the alternative isoforms of mdm2 in the damage response pathway. We have shown that mouse transcripts are structurally similar to those found in human cells. This suggests that the mouse proteins would have overlapping function with the human isoforms. Further, the predominant form of alternate mdm2 is MS2, which exhibits the exact splicing pattern of MDM2 KB2 (12) and, once again, maintains the COOH-terminal RING domain but lacks the NH₂-terminal p53-binding domain. Furthermore, we have shown that, in the mouse, MDM2 is subject to regulation by binding to and relocalization by alternative proteins that lack the p53-binding domain and nuclear localization signal. In the MDM2^alt1-transfected mouse cells, the striking increase in the protein levels of endogenous MDM2 in the cells transfected with MDM2^alt1 could be due to the up-regulation of p53 that would be expected on MDM2, preferentially binding to an alternative spliced form instead of p53 or through the putative inhibition of MDM2^alt1 of MDM2 autoubiquitination. Plainly, we have shown that mouse cells treated with UV are also characterized by alternative transcripts of mdm2, suggesting that this is an important and evolutionarily conserved mechanism for regulating the expression of mdm2 that could ultimately control the activity of p53.

The genetic studies in mouse uphold the finding that alternative splicing of mdm2 is independent of p53, as p53-null MEFs retain their ability to express the alternate forms. Additionally, ARF^−/− MEFs likewise retain their ability to express alternatively spliced forms of mdm2 in response to damage. This is consistent with the findings in human cell lines, as the MCF7 cells do not express ARF. Additionally, as with the human MDM2, alternative splicing of mouse mdm2 is not observed in response to IR.

We have found that an MDM2 family member and additional negative regulator of p53, MDM4, is alternatively spliced with the same kinetics as MDM2. A more divergent regulator of p53, Pirh2, however, is not regulated at the level of splicing in response to damage.³

The XALT2 damage-induced spliced form lacks the p53-binding domain as do the MDM2 alternative spliced forms. It is possible that this form negatively regulates full-length MDM4 in a manner similar to MDM2. Conversely, the XALT1 form of MDM4 retains only the p53-binding domain. It has previously been shown that a similar transcript (HDMX-E), which lacks exon 6 and encodes a protein only 24 amino acids longer that the predicted protein for XALT1, binds more strongly to p53 and possesses a potent p53 suppressive activity (24, 25). Thus, it seems that these two UV-inducible transcripts have antagonistic functions that may be regulated by the ratio of the two products in the cell. Interestingly, the remarkable conservation in the genomic architecture of the MDM2 and MDM4 genes and the similarities in the predominant spliced forms suggest that conserved splicing elements within the MDM2 and MDM4 genes may control the regulated splicing in response to damage (Fig. 5A). An UV-responsive splicing factor that bound to these regulatory elements would be novel and firmly place regulated splicing in the DNA damage response pathway for the first time.

Taken together, our data suggest that a protective strategy is being pursued by the induction of MDM2 and MDM4 alternative splicing that could result in the subsequent induction of the p53 pathway. Because the expression of these isoforms in tumor cells is inconsistent with their function as inhibitors of proliferation, we further hypothesize that these transcripts foster spontaneous deletion or mutation of p53 and/or arf or overexpression of MDM2 and subsequently allow for the selection of surviving immortalized cells. The proposed mechanism for MDM2^alt1 and MDM4^xalt2 regulation of p53 parallels the myc interaction with the ARF-MDM2-p53 pathway (Fig. 5B). Eischen et al. (30) have shown that transgenic mice, in which c-myc is overexpressed in B-cell progenitors, develop and die of B-cell lymphomas with a mean survival time of 6 months. They propose that the protracted latent period before the onset of the disease is reflective of the ability of c-myc to induce a “p53-dependent apoptotic program that initially protects animals against tumor formation but is disabled when overtly malignant cells emerge.” The induction of this apoptotic pathway is indicative of the resultant activation of ARF, which represses the p53 antagonistic function of mdm2 and in turn up-regulates (stabilizes) p53. Spontaneous deletion or mutation of p53 and/or arf or overexpression of MDM2 subsequently allows for the selection of surviving immortalized cells. We propose that a similar “protective” strategy is being pursued by the induction of MDM2 and MDM4 alternative transcripts in response to DNA-damaging reagents. Additionally, these data suggest that some transformed cells have a constant signal to activate p53 through sustained expression of MDM2 and MDM4 alternative transcripts.

Furthermore, this work elucidates regulated splicing as a novel mechanism by which cellular injury can control the distribution and activity of p53 within the cell. By activating alternative splicing of both MDM2 and MDM4, yet another layer of p53 regulation is initiated by the cell in response to damage. Many questions remain about the mechanism through which this alternative splicing is achieved and the true biological role these spliced forms play in the induction of the p53 pathway. A new mouse model that will help to uncover these answers is currently being developed and characterized. This mouse will be able to express only the full-length form of MDM2 and will not be regulated at the level of splicing. This new model will be an important tool to aid in the understanding of the significance of MDM2 alternative splicing and its consequences as a result of damage induction.

Grant support: NIH grant CA47296 (G. Lozano) and National Research Service Award grant CA90036 (D.S. Chandler). All sequencing was done by the M.D. Anderson Cancer Center DNA sequencing core, which is supported by NIH core grant CA16672, or by the Columbus Children's Research Institute DNA sequencing core.

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.

We thank Tomoo Iwakuma (S.S.S Cancer Center, Louisiana State University Health Science Center, New Orleans, LA) and Hugo Caldas (Center for Childhood Cancer, Columbus Children's Research Institute, Columbus, OH) for helpful discussions and comments and Dr. Andras Nagy (Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada) for the pCCALL2 plasmid.