Introduction
Oncogenes and tumor-suppressor genes are defined by their ability to transform cells upon activation (including upregulation) and inactivation, respectively. The most important requirement defining these gene classes is that the molecular changes must take place in real tumors, contributing to the initiation or maintenance of cancer, or both. SF2/ASF, one of the first characterized members of the SR protein family, meets this criterion, as Karni et al.1 now systematically demonstrate in the report on page 185 of this issue. Before this study, upregulation of SR proteins had been detected in various human tumors, but overexpression alone does not necessarily qualify a gene as an oncogene. By examining 
300 tumor samples, in the largest and most comprehensive survey yet conducted of splicing factors in human cancer, the authors show that the expression of SR proteins in general and SF2/ASF in particular is frequently elevated in tumors. In some cases, SF2/ASF overexpression results from gene amplification, one of the major modes for the activation of proto-oncogenes in human cancer. The authors next demonstrated the causal involvement of SR proteins in cellular transformation and elucidated key molecular events that lead to the tumor phenotype.
Experimental and computational studies suggest that the majority of human genes express more than one messenger RNA (mRNA) isoform via alternative splicing, which has the potential to greatly increase the functional diversity of the limited number of genes encoded in human and other mammalian genomes2. In alternative splicing, primary gene transcripts are post-transcriptionally processed to include or exclude specific RNA sequences, which may change the stability of the mRNAs or alter their coding capacities to generate protein products with different biochemical properties and thus distinct functions (Fig. 1). In several extreme situations, alternative splicing can generate gene products with opposite functions, such as those involved in apoptosis3. In general, however, we know little about how most mRNA isoforms detected contribute to specific functional consequences in biological settings and how alternative splicing may be regulated or misregulated in development and disease, despite an increasing number of examples being documented in the literature4.
Figure 1: Regulation of cellular function by alternative splicing.
Differential expression of splicing regulators, such as SR proteins and heterogeneous nuclear ribonucleoproteins, modulates alternative splicing of many cellular genes, which alters the structure of the resulting RNA as well as the function of proteins translated from alternatively spliced mRNA. Many examples of functional consequences can be found in review articles2, 3, 4 and the report by Karni et al.1. Molecular changes may collectively transform cells from one functional state to another.
Kim Caeser
Full size image (78 KB)Many disease-associated alternative splicing events have been found to result from mutations in conserved splicing signals or cis-acting regulatory elements5, but little is known about the contribution of trans-acting splicing regulators to specific disease processes. SR proteins have been extensively characterized as a class of RNA-binding proteins that function in both constitutive and regulated splicing in higher eukaryotic cells through their sequence-specific recognition of cis-acting splicing enhancers in exons and subsequent recruitment of other splicing factors to facilitate the assembly of the spliceosome6. Constitutive exons often contain multiple elements responsive to different SR proteins, which may substitute for one another in facilitating spliceosome assembly. By this means, constitutive splicing events essential for proper expression of most cellular mRNAs may be protected against variation of a given SR protein in different cell types and tissues. In contrast, exons without these features are particularly sensitive to differential expression of specific SR proteins. It is the inclusion or exclusion of those alternative exons or exonic regions that modulates the functional properties of the final gene products. Like the activities of other 'master' regulators in key biological pathways, it has long been suspected that the activity of SR proteins in alternative splicing is fundamental in modulating diverse cellular functions, and defects in these splicing factors can certainly produce a disease phenotype7. The results in the report by Karni et al.1 mimic splicing factor upregulation in human tumors, showing that SR proteins can indeed trigger cellular transformation to allow anchorage-independent cell growth in soft-agar assays and tumor formation in nude mice. Although these oncogenic assays qualify SR proteins as proto-oncogenes, it is likely that overexpression of SR proteins must combine with other crucial changes during tumorigenesis in humans, because human cells are more difficult to transform in vitro than rodent cells.
Regardless of whether SR protein overexpression is a primary trigger or a crucial contributor to tumor formation in humans, the important question is how slight overexpression of SR proteins, especially SF2/ASF, could give rise to a tumor phenotype in mammalian cells. Because SR proteins are well known for their concentration-dependent activities in alternative splicing, Karni et al.1 tested a number of known oncogenes and tumor-suppressor genes for changes in alternative splicing, which might be crucial for cells to gain advantages in growth and proliferation. Indeed, expression of a tumor-suppressor gene involved in apoptosis, BIN1, is modulated at the splicing level, and this modulation correlates with decreased levels of apoptosis in SR protein-overexpressing cells. In parallel, several kinases involved in signal transduction were also alternatively spliced in response to elevated SF2/ASF expression. These included MNK2, a kinase that phosphorylates the translation initiation factor eIF4E, and S6K1, another kinase involved in translational control in the mTOR pathway. As a result of MNK2 alternative splicing modulated by SF2/ASF overexpression, eIF4E phosphorylation was induced without the activation of upstream kinases in the MAPK pathway.
It is likely that SF2/ASF overexpression induces alternative splicing of many other genes, which together contribute to the cellular transformation phenotype. Remarkably, however, by mimicking the splicing switch of S6K1 in mouse cells, Karni et al.1 showed that one isoform (isoform-1, which encodes the known 70-kDa S6 kinase) has little oncogenic potential, whereas a novel, SF2/ASF-induced isoform (isoform-2, which encodes a 31-kDa protein with a different C terminus) is sufficient to cause a similar transformation phenotype. The splicing switch was further correlated with SF2/ASF expression in lung tumors as well as in several lung and breast cancer cell lines. These findings suggest a key contribution of the alternatively spliced S6K1 isoform to the observed cellular transformation phenotype. In light of this dramatic difference between the two S6K1 isoforms, it will be of great interest to determine in future studies how the biochemical properties of S6K1 may be altered as a consequence of the splicing switch and how the novel S6K1 isoform may perturb the mTOR pathway in cancer.
It is perhaps worth commenting on the robust experimental approach used to demonstrate the specificity of SF2/ASF in the regulation of alternative splicing and the effects of the induced mRNA isoforms on cellular transformation. The researchers combined overexpression with RNA interference-mediated knockout, instead of either one alone. The dosage effect of SR protein expression was also nicely demonstrated by small hairpin RNA-mediated knockout, which showed the reversal of the observed tumor formation phenotype, suggesting that SR proteins have an additional role in tumor maintenance. The data strongly suggest a direct effect on the observed splicing changes, although it remains possible that an overexpressed SR protein may alter the overall regulatory program in the cell by modulating the functions of other RNA-binding splicing regulators through feedback loops. SR and other splicing regulators could thereby contribute together to the complex cellular transformation phenotype.
These observations provide key molecular insights into the tumor phenotype in SF2/ASF-overexpressing cells. Although some crucial single events may be more important than others in specific biological assays, the combinatorial effects of multiple alternative splicing events may allow tumor development and selection in vivo. The new information adds to other functions previously attributed to SR proteins, especially to SF2/ASF, in the etiology of cancer. In particular, SF2/ASF has been found to control alternative splicing of the oncogene Ron to modulate cell motility, a general property of metastatic cancer cells8. Furthermore, SF2/ASF is also reported to be important in maintaining genomic stability9. As Karni et al.1 discuss, overexpression of SF2/ASF is unlikely to trigger double-stranded DNA breaks. In contrast, however, SF2/ASF knockdown has been shown to trigger such breaks9; thus, downregulation of SR proteins in some cellular contexts may also contribute to tumor development and selection. In closing, Karni et al.1 suggest that SR proteins are an important class of genetic modifiers in human diseases. Unique mRNA isoforms regulated by these proteins could thus serve as biomarkers for specific disease states10 as well as potential targets for therapeutic purposes11.