Tumorigenesis promotes Mdm4-S overexpression

Disruption of the p53 tumor suppressor pathway is a primary cause of tumorigenesis. In addition to mutation of the p53 gene itself, overexpression of major negative regulators of p53, MDM2 and MDM4, also act as drivers for tumor development. Recent studies suggest that expression of splice variants of Mdm2 and Mdm4 may be similarly involved in tumor development. In particular, multiple studies show that expression of a splice variant of MDM4, MDM4-S correlates with tumor aggressiveness and can be used as a prognostic marker in different tumor types. However, in the absence of prospective studies, it is not clear whether expression of MDM4-S in itself is oncogenic or is simply an outcome of tumorigenesis. Here we have examined the role of Mdm4-S in tumor development in a transgenic mouse model. Our results suggest that splicing of Mdm4 does not promote tumor development and does not cooperate with other oncogenic insults to alter tumor latency or aggressiveness. We conclude that Mdm4-S overexpression is a consequence of splicing defects in tumor cells rather than a cause of tumor evolution.


INTRODUCTION
Regulation of the p53 tumor suppressor and transcriptional activator is critical for normal cellular proliferation and survival. Mdm4, a homologue of Mdm2, is a bona fide negative regulator of p53 [1]. Genetic ablation of Mdm4 results in p53-dependent early embryonic lethality in mice [2][3][4], while deletion of Mdm4 in adult mouse tissues leads to subtle activation of p53 that can be potentially harnessed for therapeutic purposes [5]. Mdm4 inhibits the transcriptional activity of p53 by binding to and masking its transcriptional activation domain. In addition, Mdm4 heterodimerizes with Mdm2 through its RING domain to generate an effective E3ubiquitin ligase that degrades p53 protein during early embryogenesis [6,7].
Overexpression and amplification of MDM4 is a common theme associated with suppression of the p53 pathway in a range of human tumors [8][9][10][11][12][13]. In addition, multiple studies have highlighted the expression of MDM4 spliced variants in different tumors types [11,14]. One MDM4 spliced variant, MDM4-S that skips exon 6 and prematurely terminates in exon 7 has been the subject of close scrutiny in recent years [15]. The MDM4-S transcript potentially encodes a truncated Mdm4 protein carrying only the N-terminal p53-binding domain along with 13 novel amino acids [16]. Overexpression of MDM4-S has been linked to poor prognosis in osteosarcoma, soft tissue sarcoma, breast cancer, glioblastoma, melanoma, and chronic lymphocytic leukemia [17][18][19][20][21]. MDM4-S lacks an internal autoinhibitory sequence [22] and previous overexpression studies indicated that nuclear localized MDM4-S acts as a potent inhibitor of p53 activity and thus likely functions as an oncogene [15,16]. Recent studies however suggest that the MDM4-S transcript is susceptible to nonsense mediated decay and thus acts as a critical determinant of MDM4 expression in tumors with mutant p53 [17,18]. It is not clear why tumor cells with a defective p53 pathway would require a reduction in fulllength Mdm4 levels.
While the oncogenic potential of the full length Mdm4 protein has been clearly demonstrated by mouse studies [23], similar studies characterizing the Mdm4-S Priority Research Paper form are lacking. In the absence of prospective animal studies, it is not clear whether Mdm4-S overexpression is a cause or a consequence of tumorigenesis. To address this issue, a knock-in mouse was recently generated [24]. Unfortunately, the mouse was embryonic lethal due to excessive p53 activity. This suggests that overexpression of Mdm4-S form is not sufficient to overcome the loss of endogenous full length Mdm4. Alternatively, it is possible that loss of the C-terminus RING domain function in these mice lead to pre-natal lethality [6,7].
We recently examined MDM4-S expression in B-cell chronic lymphocytic leukemia (B-CLL, a B-cell associated malignancy) patient samples. Interestingly, over 47% of patient samples overexpressed MDM4-S mRNA compared to normal healthy volunteer controls. To gain more insight into the role of MDM4-S splice variant in B-cell malignancies, we generated a transgenic mouse wherein Mdm4-S overexpression was regulated by the Ig gene VH promoter along with the intronic enhancer and restricted to the B-cell lineage. This strategy prevented the early embryonic lethality of truncated Mdm4 expression and allowed us to investigate the significance of Mdm4-S overexpression in development of B-cell hematopoietic malignancies in mice. Herein, we present the results obtained with this mouse model and provide evidence that splicing of Mdm4 is a consequence rather than a cause of tumorigenesis.

MDM4-S is overexpressed in B-CLL patient samples
Recent studies have linked MDM4-S overexpression to poor prognosis in human cancer [17]. In order to test whether the MDM4-S form is overexpressed in B-CLL, we carried out an initial screening of 36 unselected samples obtained from patients with newly diagnosed B-CLL prior to undergoing standard chemoimmunotherapy with fludarabine, cyclophosphamide, and rituximab (i.e. FCR regimen) at the MD Anderson Cancer Center (Table 1). Overall, 47% (17/36) of patient samples showed increased expression of the MDM4 splice variant, MDM4-S to varying extents ranging from 2 fold to 42 fold as compared with B-cells obtained from normal healthy volunteers ( Figure 1A). Next, we compared the expression of full length MDM4 and the corresponding MDM4-S/MDM4 ratio in these patient samples. The MDM4-S/MDM4 ratio in these patients was also noticeably higher than normal controls ( Figure 1B). Of note, during the conduction of this project a separate study from China also reported higher levels of MDM4-S in a larger cohort of B-CLL patients correlating with poor prognosis after FCR therapy [20]. Overall, these two independent reports clearly implicate MDM4-S overexpression in the pathogenesis of B-CLL.

Generation of Mdm4-S overexpression mouse
In order to examine a possible pathogenetic role of MDM4-S overexpression in B-CLL, we cloned a murine Mdm4-S cDNA into the pBSVE6BK vector ( Figure 2A). This transgene vector harbors the Ig gene VH promoter and the intronic enhancer and has been shown to drive gene expression restricted to the B-cell lineage, thus allowing us to specifically target Mdm4-S expression in the cell of origin of B-CLL [25]. We confirmed the presence of the Mdm4-S transgene in tail snip DNA of 4 of 10 founder mice (data not shown). Based on subsequent characterization, we pursued two lines, henceforth called 2MX-S and 8MX-S, for germline transmission and followup studies.

Characterization of Mdm4-S overexpression mouse lines
As our original strategy targeted overexpression of Mdm4-S mRNA in the B-cell lineage, we first examined Mdm4-S expression in the spleen (a predominantly B-cell organ) of these mice. Compared to wild type controls, 2MX-S and 8MX-S mouse spleens showed 60-and 20fold higher expression of Mdm4-S mRNA, respectively ( Figure 2B). Next, we analyzed the mRNA expression of full length Mdm4 and the corresponding Mdm4-S/Mdm4 ratio in these two mouse lines. The Mdm4-S/Mdm4 ratio was significantly higher (19x and 13x, respectively) in these mice compared to the wild type controls. Further, we examined the Mdm4-S protein levels in the spleens of these mice. An increased abundance of Mdm4-S protein was detected in the spleens of 2MX-S and 8MX-S mice, while no difference in Mdm4 levels was observed across the different mouse genotypes ( Figure 2C). Of note, no Mdm4-S protein expression was noted in wild type mouse spleens indicating lack of detectable levels of Mdm4-S in normal animal tissues.
Mdm4-S encodes a truncated Mdm4 protein with only the N-terminal p53 binding domain intact [15]. To confirm whether the overexpressed Mdm4-S protein in our transgenic mice similarly binds to endogenous p53, we performed co-immunoprecipitation experiments. We immunoprecipitated protein lysates from spleens of wild type, 2MX-S and 8MX-S mice with an anti-p53 antibody and subjected it to immunoblotting with anti-Mdm4 antibody ( Figure 2D). As expected, we could pull down Mdm4-S from 2MX-S and 8MX-S protein lysate with a p53 antibody confirming the interaction between the two proteins, though the interaction appeared much weaker compared to Mdm4 and p53 proteins. We also examined p53 transcriptional activity in the spleens of these transgenic mice ( Figure 2E). Interestingly, mRNA expression of p53 downstream targets p21 and Puma was slightly lower in 2MX-S and 8MX-S mouse spleens compared to their wild type counterpart, but the difference was not statistically significant.
In order to determine whether overexpression of the Mdm4-S variant in B-cells leads to hematological alterations in transgenic mice, we next examined hemograms of 2MX-S and 8MX-S mice ( Figure 3A). No variation in complete blood cell count was observed in 16-20 month old 2MX-S and 8MX-S mice when compared to age matched wild-type control mice. In addition, blood smears from the transgenic mice of different ages also failed to reveal any significant morphological or quantitative differences compared to smears obtained from normal controls. No sign of malignant transformation was detected ( Figure 3B). Next, we proceeded with complete histo-pathological examination of spleens obtained from 2MX-S and 8MX-S mice of different ages. Again, no splenomegaly or other obvious signs of disease were noted ( Figure 3B). Finally, to evaluate whether overexpression of the Mdm4-S variant leads to a B-cell malignant phenotype in mice, we performed flow cytometry to  evaluate CD5 and CD19 co-expression on splenocytes. CD5:CD19 double positivity is a characteristic phenotypic feature of B-CLL cells [26]. Remarkably, no differences in double positive cell counts was observed in wild type and transgenic mouse spleens ( Figure 3C). Of note, for some unknown reasons the percentage of double positive CD5:CD19 cells in 8MX-S mice was noticeably lower than the wild type controls.

Tumorigenic profile and survival of Mdm4-S overexpressing transgenic mouse lines
To evaluate the physiological impact of Mdm4-S overexpression in mice of both transgenic lines, we generated cohorts of 2MX-S and 8MX-S mice and monitored them for tumor development and overall survival ( Figure 4A). Surprisingly, none of the transgenic mice overexpressing Mdm4-S developed a B-cell malignant phenotype. In addition, no apparent change in overall survival was noted in Mdm4-S overexpressing mouse lines. Most mice were sacrificed at the end of study (i.e. 600 days).
Recent studies have reported that increased splicing at the MDM4 locus counters the expression of full length Mdm4 [17,18]. In addition, an increase in MDM4-S/ MDM4 ratio has been proposed as a marker of poor prognosis in different human cancers [17]. In order to test whether an increase in Mdm4-S in conjunction with a corresponding decrease in full length Mdm4 promotes B-cell malignancy, we crossed Mdm4-S transgenic mice to Mdm4 Δ2 mice. Mdm4 Δ2 is a null allele previously generated in our lab [27]. We monitored a cohort of 2MX-S:Mdm4 +/ Δ2 and 8MX-S: Mdm4 +/Δ2 mice for tumor development. Interestingly, even with genetically half gene dosage of the full length Mdm4, 2MX-S:Mdm4 +/Δ2 and 8MX-S:Mdm4 +/Δ2 mice did not develop any B-cell malignant phenotype and lived a normal life span ( Figure 4B).
Lastly, we tested whether Mdm4-S overexpression cooperates with other oncogenes in tumorigenesis. To that end, we crossed Mdm4-S mice with Eµ-TCL1 transgenic mice, a prototypical B-CLL mouse model expressing the T cell leukemia-1(TCL1) transgene in B-cells that displays similar clinical and therapeutic response properties to human B-CLL [28]. We generated a cohort of 2MX-S:TCL1, and 8MX-S:TCL1 mice and investigated whether cooperation of these two genetic events leads to earlier tumorigenesis and shorter survival in these compound mice ( Figure 4C). Intriguingly, genetic combination of these two events did not alter the B-CLL latency or the overall survival of these mice. Failure to exacerbate the disease profile ruled out cooperativity between these two oncogenic insults.

Mdm4-S overexpression is a consequence of tumorigenesis
Finally, given the lack of an overt malignant phenotype, we investigated whether overexpression of Mdm4-S is a consequence rather than a cause of tumorigenic events. To that end, we isolated spleens from 3 week old p53 +/+ , p53 ̶ / ̶ , and p53 R172H/H (inheriting 2 copies of the p53R172H hotspot mutation) pre-tumorigenic mice. Separately, we also collected different types of tumors (lymphoma and osteosarcoma) that originated in p53 + / ̶ , p53 ̶ / ̶ , and p53 R172H/H mice. We isolated RNA from these normal pre-tumorigenic spleens and mouse tumors and performed RT-PCR analysis to compare the Mdm4-S expression in these biological samples ( Figure  5A). Notably, we did not observe high levels of Mdm4-S mRNA in the spleens of pre-tumor mice of any genotype ( Figure 5A, lanes 1-9). However, the Mdm4-S mRNA was prominently expressed in all 16 mouse tumor samples examined ( Figure 5A, lanes 10-25). Next, we tested whether splicing of other genes is also altered in the above mouse spleens and tumor samples. To that end we examined the expression of Fibroblast Growth Factor Receptor 1 (Fgfr1) mRNA in these biological samples. Fgfr1 produces a splice variant by exon 3 skipping in tumors. Similar to the Mdm4-S expression, very little expression of Fgfr1 splice variant was observed in pretumorigenic spleens ( Figure 5). In contrast, a noticeable expression of Fgfr1 splice variant was evident in all the murine tumor samples.
Finally, to rule out the possibility that pre-existing genomic instability in the p53 + / ̶ , p53 ̶ / ̶ , and p53 R172H/H mice promoted splicing events in the above experiment, we also screened for Mdm4-S expression in non-muscle invasive and muscle invasive bladder cancer (NMIBC and MIBC) samples that originated in wild type mice after treatment with N-Butyl-N-(4-hydroxybutyl)nitrosamine (BBN), a urothelial carcinogen. Again, real time PCR analysis indicated that increased Mdm4-S expression was confined to tumor samples ( Figure 5B). Relatively higher Mdm4-S mRNA levels were noted in MIBC samples compared to NMIBC samples, though this was not statistically significant. Altogether, these results highlight that expression of gene splice variants is a phenomenon associated with tumorigenesis but not necessarily the cause of it.

DISCUSSION
Disruption of the p53 pathway by overexpression of MDM4 is a common theme in many different types of human cancers. Recent studies have shown that the MDM4 locus produces multiple splice variants in tumors as well as in culture cells that are exposed to stress conditions [15]. In particular, one such splice variant MDM4-S, generated by exon 6 skipping correlates with poor prognosis in different types of cancers [11,17,20]. However, in the absence of prospective studies, it is not clear whether MDM4-S overexpression is a cause or consequence of tumorigenesis.
Following our initial observations that MDM4-S transcript is overexpressed in B-CLL and to directly address this question, we generated a transgenic mouse in which Mdm4-S overexpression was restricted to the B-cell lineage. This strategy allowed us to avoid the embryonic death fate and directly address the significance of Mdm4-S overexpression in B-cell associated tumorigenesis. Notably, we validated Mdm4-S expression in spleens of the transgenic mice and also confirmed the interaction between transgene derived Mdm4-S and the endogenous p53 protein. A relatively weak p53-Mdm4-S interaction along with slight decrease in basal p53 transcriptional activity was observed. Nonetheless, overexpression of Mdm4-S variant in the B-cell lineage did not result in a malignant phenotype thereby ruling out untoward suppression of the p53 pathway. Recent studies indicate that MDM4-S transcripts regulate the abundance of MDM4 protein and that the MDM4-S/MDM4 ratio correlates with tumor aggressiveness [17]. In order to test this we modulated the Mdm4-S/Mdm4 ratio by crossing Mdm4-S transgenic mice with Mdm4 Δ2 mice. Surprisingly, decreasing the levels of full-length Mdm4 by 50% and thereby increasing the Mdm4-S/Mdm4 ratio also failed to promote B-cell linked tumorigenesis. These results are similar to the previous report that showed decrease in Mdm4-S/ Mdm4 ratio had little effect on adult mouse tissues [24]. Furthermore, genetic crosses with a prototypical B-CLL mouse model also showed that Mdm4-S expression does not cooperate with other oncogenic insults such as TCL1 transgene expression to alter B-cell tumor latency in mice. Altogether, these results argue against the speculated role of Mdm4-S expression as an oncogene in B-cell lymphomagenesis.
Finally, it has been proposed that MDM4-S overexpression can serve as an effective biomarker for p53 pathway attenuation in cancers than p53 gene mutation itself [17]. This was clearly correlative in osteosarcoma and breast cancers [17,21]. However, we did not observe any correlation between MDM4-S expression and p53 mutation status in the human B-CLL samples ( Figure  1). Of note, we did not analyze expression levels of Mdm2 or other p53 inhibitors and therefore cannot rule out p53 pathway attenuation by other mechanisms. Importantly, prominent Mdm4-S expression was restricted to mouse tumors with genetic loss of p53 function and no expression was observed in normal mouse spleens ( Figure  5). Moreover, Mdm4-S protein was not detected in mouse tumors that exhibited high levels of Mdm4-S mRNA (data not shown). This is in agreement with the recent reports that suggest that the endogenous Mdm4-S transcript is susceptible to nonsense mediated decay [18,24,29].
Nonetheless, the expression of splice variant of Mdm4 and one other gene was clearly evident in a range of mouse tumor samples. Previously, genome-wide studies have revealed that large scale alterations in alternative splicing are associated with tumorigenesis [30,31] and mutations in SF3B1, a core component of spliceosome machinery correlate with disease aggressiveness and shorter survival in CLL [32]. Furthermore, it has been proposed that MDM4 alternative splicing serves as a key sensor of defects in the constitutive spliceosomal machinery that result in activation of the p53 response [29]. In keeping with this function, it is no surprise that 1). p53 is commonly lost or mutated due to selective pressure in tumors with expression of MDM4-S and 2). MDM4-S expression correlates with tumor aggressiveness and poor prognosis in different human cancers [17]. The relevance of our study is that we have demonstrated that overexpression of Mdm4-S in B-cells is not oncogenic per se and is rather a consequence of tumorigenesis, at least in CLL. Thus, our in vivo study supports the idea of using Mdm4-S mRNA levels as a possible biomarker but cautions against over-interpreting splicing data for therapeutic purposes.

Human B-CLL patient sample collection
Human samples were obtained from patients with newly diagnosed B-CLL prior to receiving standard frontline therapy with the fludarabine, cyclophosphamide, and rituximab (i.e. FCR regimen). Samples were obtained from the Department of Leukemia Tissue Bank at MD Anderson Cancer Center where they had been stored after being acquired during routine diagnostic assessments in accordance with the regulations and protocols approved by the MD Anderson Cancer Center Investigational Review Board. Signed informed consent was obtained from all patients and research was conducted in accordance with the Declaration of Helsinki. Samples were examined by a hematopathologist and graded per Rai stage. ZAP70 expression and IgVH mutation status as prognostic markers was confirmed. Chromosome 12 trisomy, 13q deletion, ATM deletion and p53 deletion were assessed by fluorescence in situ hybridization (FISH) analysis. Mdm2 SNP309 status was also determined.

Generation of transgenic Mdm4-S mice
Mouse Mdm4-S cDNA was cloned as an EcoRV-Sal1 fragment into the pBSVE6BK vector. Linearized vector was injected into blastocysts to generate chimeric mice by the Genetically Engineered Mouse Facility at the MD Anderson Cancer Center. DNA from mouse tail snips was PCR amplified with vector specific forward primer (acccagatgtcccttcttctccag) and Mdm4-S specific reverse primer (gactcgagtcagttctttttctgggattg) to confirm the presence of the transgene. All mouse studies were approved by MD Anderson Cancer Center IACUC. TCL1 mice were a gift from Dr. Croce (Ohio State University, Columbus, OH). Mdm4 Δ2 mice were previously generated in our lab [27].

RNA isolation and real time PCR
RNA was isolated from mouse spleen or human B-CLL patient lymphocytes using TRIzol (Thermo Fisher Scientific). After DNase1 treatment and phenolchloroform extraction, RNA was precipitated and suspended in DNase/RNase free water. One microgram of RNA was used for first strand synthesis (First-Strand cDNA Synthesis Kit, GE Life Sciences). Real time PCR was carried out as previously described [6].