|Home | About | Journals | Submit | Contact Us | Français|
Tumor cells display a different profile of gene expression than their normal counterparts. Perturbations in the levels of cellular splicing factors can alter gene expression, potentially leading to tumorigenesis. We found that splicing factor SRp20 (SFRS3) is highly expressed in cancers. SRp20 regulated the expression of Forkhead box transcription factor M1 (FoxM1) and two of its transcriptional targets, PLK1 and Cdc25B, and controlled cell cycle progression and proliferation. Cancer cells with RNAi-mediated reduction of SRp20 expression exhibited G2/M arrest, growth retardation, and apoptosis. Increased SRp20 expression in rodent fibroblasts promoted immortal cell growth and transformation. More importantly, we found that SRp20 promoted tumor induction and the maintenance of tumor growth in nude mice and rendered immortal rodent fibroblasts tumorigenic. Collectively, these results suggest that increased SRp20 expression in tumor cells is a critical step for tumor initiation, progression, and maintenance.
Alternative RNA splicing, a principal molecular event for the gene expression of approximately 70% of all human genes 1,2, increases the coding capacity of the human genome by producing different isoforms from a single pre-mRNA molecule. The regulation of alternative splicing involves interactions between cellular splicing factors and RNA sequences in the pre-mRNA 3,4 and can easily be perturbed by relatively small changes in the levels of splicing factors 5,6. Although alternative splicing events have recently emerged as an important focus in molecular and clinical oncology 7-9, the contribution of alternative splicing to cancer development is poorly understood.
SRp20, recently renamed as SFRS3 10, is a splicing factor that affects alternative splicing by interacting with RNA cis-elements in a concentration- and cell differentiation-dependent manner 11,12. It is the smallest member of the SR protein family 13. In addition to its regulation of RNA splicing, SRp20 plays important roles in cellular functions including termination of transcription 14, alternative RNA polyadenylation 15, RNA export 16,17, and protein translation 18. Mouse embryos lacking SRp20 do not form a blastocyst 19. SRp20 expression is higher than normal in ovarian cancers 20; however, the effect of this increased expression is unclear. Overexpressed SRp20 might alter the RNA splicing and other events of many genes in mammalian cells, thereby substantially affecting the expression levels of various protein isoforms 21. Here we provide evidence that SRp20 is overexpressed in many cancer types and that the increase in SRp20 is essential for cancer cell survival and oncogenesis. In addition, we found that SRp20 is critical for controlling the cell G2/M phase transition and for preventing cell apoptosis.
A T7-SRp20 expression vector (plasmid pJR17) was constructed by swapping the T7-SRp20 coding region from a pCG-T7-SRp20 expression vector (Tom Misteli of NCI and Javier Caceres of Edinburgh) into the pRevTRE vector to put T7-SRp20 under the control of tetracycline.
Immortal rodent fibroblast NIH 3T3 and MEF 3T3 tet-off cells (Clontech) and human C33A, 786-O, U2OS, HeLa, CaSki, WI-38, and MRC-5 cells were grown in Dulbecco's modified Eagle medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS) or calf serum (HyClone), 2 mM L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. The B-cell lymphoma-derived cell lines BCBL-1, JSC-1, and SUDHL-6 were maintained in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS. Primary human bronchial epithelial cells (HBEpiC) and primary human renal epithelial cells (HREC) were obtained and grown in medium from ScienCell Research Laboratories. HBEpiC were grown in bronchial epithelial cell medium, and HREC were grown in DMEM supplemented with 10% FBS and Epithelial Cell Growth Supplement (ScienCell Research Laboratories). Primary newborn human foreskin keratinocytes (HFKn) were purchased from Invitrogen and grown in calcium-free Medium 154CF plus human keratinocyte growth supplement (Invitrogen). Peripheral blood mononuclear cells were obtained from healthy blood donors at the NIH Clinical Center blood bank. MEF 3T3 tet-off cells were transfected with plasmid pJR17 (T7-SRp20) by Lipofectamine 2000 (Invitrogen), and selected with 200 μg/mL hygromycin and 1 μg/mL doxycycline for stable transfection. A cell line stably transfected with the empty vector, pRevTRE, was also established as a control. To express T7-SRp20, the stably transfected MEF3T3 tet-off cells were grown in doxycycline-free DMEM growth medium.
Tissue lysates of various pairs of tumor and matched normal tissues from different organs were purchased from Protein Biotechnologies.
Synthetic, double-stranded siRNAs were obtained from Dharmacon, Inc. SRp20 and hnRNP U siRNAs were purchased as an siGenome SMARTpool (human SRp20, cat. No. M-030081-00; human hnRNP U, cat. No. J-013501-00). Human YB-1 siRNA is a mixture of siRNAs 393, 394, and oJR1, as described 11. The nonspecific (NS) siRNA has 52% GC content (cat. No. D-001206-08-20). The SRp20 siRNA s12732, targeting a splice junction of SRp20 exon 2 and exon 3, was purchased from Ambion and SRp20 siRNA D-03 was obtained separately from Dharmacon. RNAi was conducted by two or three separate transfections, at intervals of 48 h, with 10 nM (Ambion) or 40 nM (Dharmacon) siRNA in the presence of Lipofectamine 2000. BCBL-1 and JSC-1 lymphoma cell lines were transfected with 40 nM siRNA using the HiPerfect transfection reagent (Qiagen) by three separate transfections in accordance with the instructions of the manufacturer. The cells with knocked-down SRp20, YB-1, or hnRNP U expression were then analyzed for cell number by trypan blue exclusion, for cell cycle by flow cytometry, for RNA splicing by RT-PCR, for protein expression by Western blotting, and for tumor induction by nude mouse injection.
HEKn cells at 2 × 105 cells per well in 12-well plates were transfected 4 h after passage 3 (day 0) with 40 nM SRp20 siRNA or NS siRNA by using Lipofectamine 2000. Cells were transfected again on days 2 and 4 without passage and were counted on day 6. HBEpiC cells at 2.5 × 105 cells per well in 6-well plates were transfected on days 1 and 3 with siRNA as described above for HFKn and were counted on day 5. HREC cells at 2 × 104 cells per well in 24-well plates were transfected on days 1 and 3 with siRNAs as described above for HFKn and were counted on day 5.
Protein samples in 2× SDS sample buffer were denatured by boiling for 5 min, separated by NuPAGE Bis-Tris gel electrophoresis (Invitrogen), transferred onto a nitrocellulose membrane, and blotted with the following antibodies: mouse monoclonal antibodies against SRp20 (7B4, American Type Culture Collection), beta-tubulin (BD PharMingen), hnRNP K (Santa Cruz Biotech), and PARP (Calbiochem), or rabbit antibodies against PLK1 (Millipore/Upstate), N-terminal FoxM1 (K19), and Cdc25B (Santa Cruz Biotech).
Total cell RNA was prepared from the cells using TRIzol (Invitrogen) following the manufacturer's instructions. After DNase I treatment, 1 μg of RNA was reverse transcribed at 42oC using random hexamers and then amplified using the following gene-specific primer pairs: oJR85 (5'-CGGCTGCCCTACCTACGGA-3') and oJR87 (5'-GGGAGGGCAGCTATTAGGA-3') for human PLK1; oJR104 (5'-CGGCTGCCCTACCTACGAACG-3') and oJR105 (5'- GGCACCAGTCCGAAGGAGAGA-3') for mouse PLK1; oJR111 (5'-TGAAGCCACTGCTACCACG-3') and oJR112 (5'-AGGGCTCTCCACTTTGATGG-3') for human FoxM1; oJR113 (5'-TCGGCCCCGCGTGGAGCAGAC-3') and oJR114 (5'-TAACCCGATTCTGCTCCAGGTGAC-3') for mouse FoxM1; oJR100 (5'- GGGCAAGTTCAGCAACATCGTGGA-3') and oJR101 (5'-GTAGCCGCCTTTCAGGATATACATC-3') for both human and mouse Cdc25B 22.
Total RNA purified from cells with or without SRp20 knockdown was quantified by qRT-PCR by using Applied Biosystems TaqMan probes for human PLK1 (#HS00983227) and 18S rRNA (#4333760F) in accordance with the manufacturer's instructions. qRT-PCR was performed in a Cepheid Smart cycler. Briefly, 1 μg of total RNA treated with RNase-free DNase I was reverse transcribed at 42oC using random hexamers, followed by a 20-μL PCR reaction that included 2 μL RT product, 1 μL TaqMan Gene Expression Assays (20×), and 10 μL TaqMan Universal PCR Master mix (2×). The PCR reactions were performed at 50̊C for 2 min and 95̊C for 10 min, followed by 40 cycles of 95̊C for 15 s and 60̊C for 10 min. The relative expression level of PLK1 in each sample was calculated by the 2-ΔΔCT or 2-ΔCT method 23 after it was normalized to 18S rRNA.
Cervical, soft tissue, and epithelial tumors and normal tissue sections were purchased from US BioMax. Immunohistochemistry was performed with a Vectastain ABC kit (Vector Laboratories) in accordance with the manufacturer's protocol. Sections were deparaffinized, rehydrated, and microwaved for 15 min in the presence of 1× Antigen Retrieval Citra Plus buffer (BioGenex). Endogenous peroxidases were quenched using 3% hydrogen peroxide for 15 min. Sections were incubated with an anti-SRp20 7B4 antibody overnight at 4̊C, followed by a secondary antibody for 30 min and ABC reagent for 30 min. The specific signal was developed by using a DAB substrate kit (Vector Laboratories).
NIH 3T3 cells transiently transfected with plasmid pT7-SRp20 (pJR17) or empty vector pFLAG-CMV5.1, or MEF 3T3 tet-off cells stably transfected with plasmid pJR17 or empty vector pRevTRE, were harvested, adjusted to 1 × 104 cells in DMEM containing 0.35% agar and 10% doxycycline-free FBS, and then laid onto a bottom layer containing 0.5% agar and 10% FBS in DMEM in 6-well plates. The plates were stained 3-4 weeks later with 0.005% crystal violet for colony counting.
For tumor induction with HeLa cells, cells transfected twice with SRp20 siRNA or nonspecific siRNAs were collected 24 h after the second transfection and implanted by dorsal subcutaneous inoculation of 1 × 106 cells into both sides of nude mice, with 10 mice in each group. Tumor sizes were measured at 14, 17, and 19 days after implantation. Tumor weight was recorded when the animals were sacrificed on day 22.
For tumor induction with MEF/3T3 tet-off cells, 3 × 106 cells stably transfected with a T7-SRp20 vector or with the empty vector (pRevTRE) were grown in doxycycline-free medium and implanted as described above. Tumor sizes were measured every 4 to 5 days, and tumor weight was recorded when the animals were sacrificed on day 50.
Cell cycle distribution was determined by flow cytometry. U2OS or HeLa cells transfected twice with SRp20 siRNA or NS siRNA at an interval of 48 h were trypsinized 96 h after the first siRNA transfection, washed twice with PBS, resuspended in Vindelov's propidium iodide buffer, and analyzed with a CYAN MLE cytometer (Dako-Cytomation). Ten thousand events were collected per sample, and data were analyzed with Modfit LT software (Verity Software House).
To analyze the synchronized cells for cell cycle progression, MEF 3T3 tet-off cells stably transfected with T7-SRp20x were seeded in a 6-well plate at 5 × 104 cells per well in DMEM containing 10% doxycycline-free FBS overnight, synchronized by starvation for 48 h in DMEM containing 0.1% FBS, and then collected at the indicated times after serum stimulation (with 10% FBS) for cell cycle analysis by flow cytometry as described above.
The Oncomine cancer microarray database [http://www.oncomine.com; 24] was used to analyze expression profiles of SRp20 in a variety of human cancerous and normal tissues and in association with tumor progression (grade) and 5-year survival rate. Statistics from individual studies were also obtained from the Oncomine cancer database (January 15 or April 15, 2010, version) and were combined for a Fisher's meta-analysis. Overexpression of SRp20 was defined as significantly higher expression (P < 0.05) in tumor tissues than in the corresponding normal tissues, in high-grade tumors than in low-grade tumors, or in shorter-living (<5 years) cancer patients than in longer-living (>5 years) cancer patients.
Southern blot analysis was conducted by using ~5 μg of EcoRI-digested genomic DNA extracted from paired normal and cancerous lung tissues (BioChain Institute, Hayward, CA) and hybridized with an SRp20 DNA probe, a 649-bp PCR fragment amplified with a 5' primer (oJR53, 5'-AAGCCGTCCCGATCCTTCTC-3') and a 3' primer (oJR57, 5'-GACTGCTTGTTCAACTATAGCTGCA-3') from HeLa genomic DNA and randomly labeled with 32P. The blot was reprobed for cyclophilin as a loading control by using a cyclophilin probe, a 660-bp fragment derived from a Hind III-digested 1.1-kb PCR product amplified with a 5' primer (oSB21, 5'-CCAAAGCATTGTACCGCAGAG-3') and a 3' primer (oSB22, 5'-TTGCATATACTGCCTTCTCTTTATC-3') from HeLa genomic DNA and randomly labeled with 32P.
For semi-quantitative PCR analysis of SRp20 gene amplification in paired normal and cancerous cervical or lung tissues, genomic DNA isolated from the cervical or lung tissues was serially diluted and analyzed by PCR by using a primer pair of oJR56 (5'-TCTCTTGAAACAGTGACACAAAGGTG-3') and oJR57 for SRp20 gene detection. Cyclophilin PCR with a primer pair of oSB21 and oSB22 served as loading control.
Statistical data for the paired microarray datasets in Fig. Fig.3,3, were obtained directly from the Oncomine cancer database (www.oncomine.com). Statistics from individual studies with significantly higher expression of SRp20 (P < 0.05) in tumor tissues than in the corresponding normal tissues were also obtained from the Oncomine cancer database (January 15 or April 15, 2010, version) and were combined for Fisher's meta-analysis. All two-group statistical comparisons of means in Fig. Fig.66 and Fig. Fig.1010 were calculated with two-tailed student's t test using Excel (Microsoft).
In looking at the role of SRp20 in human papillomavirus (HPV) RNA splicing 11, we found a remarkable increase of SRp20 expression in cervical cancer tissues (Fig. (Fig.1A).1A). However, this increase was not limited to cancers caused by HPV infection. We also observed variable increases of SRp20 expression in cancers of the lung, breast, stomach, skin, bladder, colon, liver, thyroid, and kidney (Fig. (Fig.1B),1B), as well as in B-cell lymphoma cells (JSC-1 [KSHV+/EBV+], BCBL1 [KSHV+], and SUDHL-6; Fig. Fig.11C).
Tissue-array immunohistochemistry demonstrated increased expression of SRp20 not only in epithelial carcinomas (Fig. (Fig.2),2), but also in mesenchymal tissue-derived tumors, including rhadbomyosarcoma, hemangioendothelioma, hemangiopericytoma, neurofibroma, neurilemmoma, liposarcoma, leiomyosarcoma, histiocytoma, and synovial sarcoma (Supplementary information, Fig. Fig.S1).S1). By searching the Oncomine cancer microarray database (http://www.oncomine.com), we found a significant increase (P < 0.05) of SRp20 expression in tumor tissues over the expression in corresponding normal tissues in 96 of 190 studies. Fisher's meta analysis indicated that the observed increase in those paired studies was significant (P <0.001). We also found that the increased SRp20 expression correlated with breast cancer progression in 13 of 26 studies (P<0.001) as represented in Fig. Fig.3A3A 25,26, and with 5-year overall survival in 3 of 6 studies (P = 0.001), as represented in Fig. Fig.3B3B 27,28.
As the gene SFRS3 which encodes SRp20 is located on chromosome 6p21, a common region of DNA amplification seen in many cancers 29, we examined whether gene amplification would be a cause for increased SRp20 expression in cancer tissues. As shown in Fig. Fig.4,4, we verified SFRS3 gene amplification in lung cancer by Southern blotting and semi-quantitative PCR and in cervical cancers by semi-quantitative PCR, demonstrating that SFRS3 gene amplification could be a cause of increased SRp20 expression in at least a subset of these cancers.
Increased SRp20 could potentially induce the production of mRNA isoforms that encode proteins favoring oncogenesis. We therefore hypothesized that increased SRp20 expression might contribute to maintenance of the carcinogenic phenotype, rather than being a consequence of it. To test this possibility, we examined whether reduced SRp20 expression would affect cancer cell growth. Each of seven cancer cell lines (U2OS, HeLa [HPV18+], CaSki [HPV16+], C33A, 786-O, JSC-1, and BCBL-1) exhibited reduced proliferation after SRp20 expression was knocked down by an siRNA pool targeting different regions of SRp20 mRNA (Fig. (Fig.5A-C,5A-C, Supplementary information, Fig. Fig.S2A-E).S2A-E). To exclude any possible off-target effect of the pooled SRp20 siRNAs, we performed similar experiments in U2OS and HeLa cells by using a separate siRNA, siRNA s12732, that targets a splice junction of SRp20 exon 2 and exon 3 (Fig. (Fig.5A)5A) and observed the same results (Fig. (Fig.5D-E).5D-E). Using siRNA D-03 alone in U2OS cells also gave the same result (Supplementary information, Fig. Fig.S2F).S2F). By contrast, knockdown of YB-1 or hnRNP K had no effect on HeLa cell growth (Supplementary information, Fig. Fig.S2G).S2G). Further studies showed that knockdown of SRp20 expression in U2OS cells led to cell apoptosis, as indicated by apoptotic cleavage of PARP (Fig. (Fig.5F).5F). These data indicate that increased SRp20 expression is necessary for the indefinite growth of cancer cells and prevents cancer cell apoptosis.
WI-38 and MRC-5 cells, two human diploid lung fibroblast cell lines with a finite lifetime of about 50 population doublings, naturally express much less SRp20 than U2OS and HeLa cells (Fig. (Fig.66A).
When SRp20 was transiently introduced into WI-38 cells, the cell growth increased (Fig. (Fig.6B).6B). Interestingly, human primary epithelial cells (human newborn foreskin keratinocytes [HFKn], human bronchial epithelial cells [HBEpiC], and human renal epithelial cells [HREC]), which have limited doubling times in culture, also express less SRp20 than cancer cell lines (Fig. (Fig.6C).6C). Knocking down SRp20 expression in these cells also slowed their growth (Fig. (Fig.6D-F).6D-F). Altogether, these data indicate that SRp20 is also essential for normal cell proliferation.
Flow cytometry showed that both U2OS and HeLa cells with siRNA-mediated reduction of SRp20 displayed prominent G2/M arrest (Fig. (Fig.7A),7A), accompanied by increased expression of FoxM1a and decreased expression of FoxM1b-c, Cdc25B 30, and polo-like kinase-1 (PLK1; 31 at the mRNA level (Fig. (Fig.7B,7B, Supplementary information, Fig. Fig.S3).S3). We noticed that U2OS cells which contain a functional G1 check-point 32, but not HeLa cells which lack a functional G1 check-point due to HPV18 E7 degradation of pRb 33, also appeared a slight increase of cell number at S phase after knocking down SRp20 expression (Fig. (Fig.7A).7A). Western blotting analyses showed reduced protein expression of FoxM1b-c, Cdc25B, and PLK1 (Fig. (Fig.7C),7C), which are all involved in the G2/M transition 34-36. Because the physiological significance of FoxM1a is not clear and it is not generally detectable by Western blot 37,38, we conclude that the increased expression of SRp20 in cancer cells promotes cell cycle progression by affecting the expression of G2/M transition regulators, thus contributing to the indefinite growth of the tumor cell lines in monolayer culture.
When ectopically expressed in immortal NIH 3T3 fibroblasts in the presence of endogenous SRp20, T7-tagged human SRp20 (T7-SRp20) conferred a substantial growth advantage to cells in monolayer culture (Fig. (Fig.8A)8A) and anchorage-independent growth to cells in soft agar (Fig. (Fig.8B).8B). This experiment was repeated in another immortal cell line, Swiss MEF (mouse embryonic fibroblast) 3T3 cells. The cells with transient (Fig. (Fig.8C)8C) or stable expression (Fig. (Fig.8D)8D) of T7-SRp20 displayed a two-fold increase in growth rate; colony formation in soft agar also increased (Fig. (Fig.88E-F).
Knockdown of SRp20 expression in cancer cells reduced the expression of the cell cycle regulators FoxM1, Cdc25B, and PLK1 and prevented cell cycle progression through G2/M phase (Fig. (Fig.7);7); in contrast, expression of T7-SRp20 in stably transfected MEF 3T3 cells in the presence of endogenous SRp20 enhanced the expression of FoxM1, Cdc25B, and PLK1 both at the RNA (Fig. (Fig.8G)8G) and protein (Fig. (Fig.8H)8H) levels and accelerated cell cycle progression of synchronized MEF 3T3 tet-off cells. As shown in Fig. Fig.9,9, overexpression of T7-SRp20 in synchronized MEF 3T3 tet-off cells pushed the cells quickly go through G2/M phase (compare cell numbers at G2/M phase from 7 h to 10 h) and accelerated the cell cycle transition from G0/G1 to S phase (compare cell numbers at G0/G1 and S phase from 10 h to 15h), over the vector controls.
Given the cell growth and transformation potential of SRp20, we conducted a series of nude mouse studies on SRp20 oncogenesis. Much lower tumor induction was seen when HeLa cells with siRNA-reduced SRp20 expression were implanted into nude mice than when HeLa cells receiving a nonspecific siRNA were implanted (Fig. (Fig.10A-C).10A-C). When implanted into nude mice, MEF 3T3 cells with stable T7-SRp20 expression in the presence of endogenous SRp20 exhibited large tumors, whereas the cells stably transfected with the control vector were not competent for tumorigenesis (Fig. (Fig.10D-F).10D-F). Together, these data indicate that increased SRp20 expression in immortalized, untransformed mammalian cells is also oncogenic in nude mice.
Although various efforts have been made to understand the molecular basis of tumorigenesis, the mechanisms that lead to tumor-specific gene expression remain largely unknown. Altered expression of splicing factors had been described in various tumor types 9,20,39-41, but the effect of increased expression of splicing factors on tumorigenesis remained to be investigated. To our knowledge, there has been no report on the role of SRp20 in the development of cancer. In this study, we demonstrated a direct cause-effect relationship between increased SRp20 expression and tumor formation with the following evidence: First, all cancer types examined exhibited increased SRp20 expression. SRp20 expression in breast cancer tissues increased with cancer progression and correlated with patients' 5-year survival. Second, cancer cells with reduced SRp20 expression appear to grow slowly, undergo apoptosis as reported while our manuscript was in preparation 42, and are less tumorigenic in nude mice. Third, overexpression of SRp20 in immortal 3T3 cells caused cell transformation and induced tumor formation in nude mice. We characterize SRp20 as a novel proto-oncogene because SRp20 is a normal cellular gene that, when expressed at its physiological level, is essential for normal cell proliferation in cultures (Fig. (Fig.6).6). The cells in the basal layers of cervix and esophagus, which are dividing cells, appear to express SRp20 relatively more than non-dividing, terminally differentiated cells (Fig. (Fig.2).2). Together with the finding that splicing factor SF2/ASF is overexpressed in lung and colon cancers and is oncogenic in nude mice 9, our data provide compelling evidence that altered expression of SR proteins is an important contributor to the development of cancer.
SRp20 that becomes oncogenic appears only when its expression from chromosome 6p is increased, presumably by gene amplification. This chromosome region bearing the SFRS3 (SRp20) gene is commonly amplified in many cancers 29. Our preliminary investigation has confirmed that gene amplification could be a cause of SRp20 overexpression in lung and cervical cancers (Fig. (Fig.4).4). Other oncogenic signaling pathways might also trigger SRp20 expression. One such pathway might be the Wnt signaling pathway, which is implicated in driving the formation of various human cancers and recruits cytosolic β-catenin as a transcriptional coactivator 43 to transactivate SRp20 expression 44. Although the specific cause of increased SRp20 expression in human cancers remains to be investigated, and there may be several possible causes, SFRS3 gene amplification in chromosome 6p as reported in many other studies 29 may lead to increased SRp20 expression in at least a subset of these cancers.
The dysregulated expression of splicing factors can change general RNA splicing and other events essential for the proper expression of the targeted genes, consequently causing diseases and tumor formation 8,45-47. In this report, we identified cellular FoxM1, PLK1, and Cdc25B as three prominent targets of SRp20 in the regulation of cell proliferation and oncogenesis (Fig. (Fig.10G).10G). FoxM1, known previously as HFH-11, WIN, MPP-2, FKHL-16, or Trident 48, is a transcription factor of the Forkhead family that regulates expression of PLK1, Cdc25B, and CENP-F 34,49-51. FoxM1 is expressed in proliferating cells in a cell cycle-dependent manner 37,52-54 and plays crucial roles in the G2/M transition 49,50 and in chromosome stability and segregation during mitosis 34. The gene encoding FoxM1 on chromosome 12p13.3 produces a primary transcript that can be alternatively spliced to create three RNA species by inclusion (FoxM1a and 1c) or exclusion (FoxM1b) of exon 6 and inclusion (FoxM1a) or exclusion (FoxM1b and 1c) of exon 9 (Supplementary information, Fig. Fig.S3);S3); these three RNA species produce three protein isoforms, FoxM1a, 1b, and 1c, with only FoxM1b and 1c being transcriptionally active 48. Two FoxM1 targets, PLK1 and Cdc25B, are crucial for entry of the cell cycle into mitosis 31,55,56, are all overexpressed in many cancers and oncogenic 48,55,57,58 and were all expressed at reduced levels in U2OS and HeLa cells with SRp20 knockdown, causing the cells to arrest at G2/M. Because PLK1-dependent phosphorylation of FoxM1 is required for G2/M progression 59, reduced PLK1 expression in the cells would result in nonfunctional FoxM1, thereby exaggerating the FoxM1 deficiency caused by SRp20 knockdown. In contrast, overexpression of SRp20 in 3T3 cells stimulated the expression of FoxM1 and its targets PLK1 and Cdc25B and promoted cell proliferation and tumorigenesis.
Although the mechanism by which SRp20 promotes FoxM1 expression remains to be determined, we assume that SRp20 regulation of FoxM1 expression is both transcriptional and posttranscriptional. Because SP1 binding to the FoxM1 promoter regulates FoxM1 expression 60, downregulation of SP1 expression in cells with reduced SRp20 levels 11 may be responsible for FoxM1 reduction at the transcriptional level. Human FoxM1 exon 9 is not conserved in the mouse genome, and human FoxM1a with the exon 9 inclusion does not encode a functional FoxM1a protein; the enhancement of mouse FoxM1 expression by SRp20 therefore presumably occurs by transcription through an indirect mechanism. On the other hand, the enhancement of human FoxM1 expression by SRp20 might occur through the regulation of RNA splicing, most likely by promoting the exclusion of FoxM1 exon 9. Nevertheless, the effects of SRp20 on FoxM1, PLK1, and Cdc25B expression are one mechanism by which SRp20 regulates cell proliferation and oncogenesis.
This research was supported by the Intramural Research Program of the Center for Cancer Research, NCI, NIH. We thank Javier Caceres and Tom Misteli for providing T7-SRp20 expression vector, Robert Yarchoan, Douglas Lowy, and Adrian Krainer for their critical comments and encouragement in the course of this study and for their critical reading of the manuscript; Curtis Harris and Izumi Horikawa for their critical reading of the manuscript; and Calvin Chan for his initial analysis of the Oncomine database.