During early mammalian development embryonic cells proliferate rapidly, driven by the expression of multiple genes that control the cell cycle.1 Proliferation subsequently subsides and eventually stops in most differentiated cells, as expression of the genes is repressed. Many of the genes involved in cell proliferation are proto-oncogenes, which are repressed by a tumor-suppressor protein (TSP) that binds to the genes and blocks transcription.2, 3 Oncogenesis begins with the transformation of a normal cell into a progenitor cell for a tumorigenic clone. Transformation could occur because of failure to repress transcription of proto-oncogenes in a proliferating cell, or reversal of proto-oncogene repression in a differentiated cell. The cells in an expanding tumor clone are genetically unstable, accumulating mutations in tumor-suppressor genes that inactivate or delete the encoded TSPs, and/or mutations in proto-oncogenes that generate constitutively expressed oncogenes.4, 5 The association of these mutations with cancer and their key role in tumor growth and dissemination are firmly established, but association does not necessarily indicate causation; mutations could be a consequence rather than a cause of transformation. A mechanism of transformation must account for the coordinate regulation of more than 100 genes that function in cell proliferation,1 which is unlikely to result from random mutations. Here we describe a mechanism for reversible regulation of gene transcription that involves the binding of RNA to a TSP, releasing the TSP from a gene and reversing repression.6, 7 This mechanism appears to function in embryogenesis and steroidogenesis as well as oncogenesis.
Role of TSP-binding RNA in oncogenesis
The TSP in our first studies was PSF,8 which belongs to a family of conserved mammalian proteins containing a DNA-binding domain (DBD) and RNA-binding domains (RBDs). This family includes other TSP such as p53,9-11 SAFB112 and RBM5/H37/LUCA15.13, 14
The PSF protein was isolated from spliceosomes, and was shown to function in RNA splicing mediated by its RBDs.8 Later studies showed that the DBD of PSF binds to P450scc, the first gene in the steroidogenic pathway, repressing transcription of P450scc and blocking steroidogenesis.15 Further studies identified additional genes repressed by PSF, including numerous proto-oncogenes, and showed that PSF suppresses proliferation and colony formation of tumor cells,7 indicating that PSF functions as a TSP. The presence of RBDs in a TSP is unexpected, because repression of transcription should require only a DBD. An explanation for the linkage of the RBDs and DBD in PSF was provided by the discovery of a novel regulatory function of the RBDs, which involves binding of a RNA to form a PSF/RNA complex that dissociates from a repressed gene and reverses repression.6, 7, 16 A similar function was described for the RBDs of p53, which form a complex with RNA that disrupts oligomerization and specific binding to DNA.10, 11 The PSF-binding RNA that have been identified include a mouse retroelement RNA from the VL30 family,16 the human retroelement RNAs L1PA16, MER65B and MER11C, and the human RNAs MALAT-1, OTP, Corf11 and HNI (Xu Song et al., in preparation). Some of the PSF-binding RNA are ncRNA, and others involve noncoding regions of coding RNA. Expression of mouse PSF-binding RNA is associated with mouse tumors,17-20 and expression of human PSF-binding RNAs is associated with human tumors,19, 21-25 consistent with a role of the PSF/RNA regulatory mechanism in mammalian oncogenesis. Additional RNAs that bind to PSF, and RNAs that bind to other TSPs, probably remain to be identified.
Many of the endogenous TSP-binding RNAs are transcribed from retroelement genes derived from earlier retroviral infections, comprising almost 50% of the DNA in mammalian genomes.26 Although a retroviral infection and integration into the germline was a rare event in a population, all species examined contain multiple copies of retroelement genes, indicating that the genes provided a selective advantage to the species. One advantage resulted from retrotranspostion of retroelement genes to new genomic sites, generating cis-acting mutations that altered expression and function of targeted genes and expanded the genetic repertoire for evolutionary selection.26 The frequency of retrotransposition has been reduced by mutations that inactivate retroelement genes and by micro RNA (miRNA) that inactivate retroelement transcripts.27 Some retroelement genes are transcribed with temporal and spatial specificity,28-31 and the retroelement RNA, although partly or completely deficient in coding potential, can function in trans to regulate gene transcription.6, 7, 16
The ras family of oncogenes is expressed in many types of human cancer and is considered to have an important role in oncogenesis that is not fully understood.32 It was reported about 20 years ago that transfection of mouse NIH 3T3 cells with the rash oncogene under control of the MMTV promoter resulted in steroid-induced expression of the ras p21 oncoprotein and cell transformation, as expected, and also in expression of VL30 RNA, which was not expected.33 The VL30 RNA contains PSF-binding sequences, suggesting that a function of the rasH oncogene in mouse cells is to activate expression of VL30 RNA, which can initiate transformation by binding to PSF. ras oncogenes could have a similar function in the transformation of human cells, by activating expression of a human TSP-binding RNA.
Most cancers examined have reduced levels of miRNA,34 resulting from a defect in miRNA processing.35 These important findings suggest that miRNA can suppress transformation of normal cells, presumably by inactivating RNA that promote transformation, and that reduction of miRNA levels increases expression of proto-oncogenes. One example involves let-7 miRNA that inactivates ras mRNA36: the level oflet-7 miRNA is reduced in human tumors,34 resulting in increased expression of rasoncogenes. Another example involves miRNA that inactivate retroelement RNA,27 probably including TSP-binding RNA: reduced levels of the miRNA in tumors could increase the levels of TSP-binding RNA that reverse repression of proto-oncogenes.6, 7, 16 It was reported recently that the RNA genome of the hepatitis delta virus binds specifically to PSF protein in HeLa cells.37 This seminal discovery suggests that the RNA of an oncogenic RNA virus, in addition to endogenous TSP-binding RNA, could promote oncogenesis is by binding to a TSP and reversing repression of proto-oncogenes.
We propose a model of oncogenesis based on the role of TSP-binding RNA, usually a ncRNA, in reversing repression of proto-oncogenes by a TSP (Fig. 1).41 During early development embryonic cells express high levels of TSP-binding RNA that drive cell proliferation by preventing TSP-mediated repression of proto-oncogenes. The level of TSP binding RNA subsequently decreases in most differentiating cells, enabling a TSP to repress proto-oncogenes and stop cell proliferation. A progenitor tumor cell can form when the level of TSP-binding RNA fails to decrease in a proliferating cell or increases in a differentiated cell. The resulting clone of proliferating tumor cells is genetically unstable, accumulating mutations that promote tumor dissemination and metastasis. The TSPs in this model belong to a family of transcriptional repressors containing a DBD and RBDs, which includes PSF, p53, SAFB1 and RBM5/H37/LUCA15. The level of a TSP is assumed to remain relatively constant, while the level of TSP-binding RNA varies with stage and tissue specificity. The level of TSP-binding RNA can increase in various ways, including exposure of cells to hormones39 or anoxia,42 expression of a ras oncogene,33 and probably a decrease in the levels of miRNA.27, 35 The elevated level of TSP-binding RNA in tumor cells suggests that the RNA provide a potential diagnostic and therapeutic target for cancer.
Role of TSP-binding RNA in embryogenesis and steroidogenesis
In addition to a pathological function in tumorigenesis, TSP-binding RNA appear also to have normal functions in promoting embryonic cell proliferation and steroidogenesis, as follows.
Mouse oocytes and embryonic cells express high levels of retroelement RNA, including VL30 (Xu Song et al., unpublished), Etn,29 EC-1,27 L1,28, 31 IAP31 and MT,31 and inactivating L1 RNA in mouse embryos aborts development.25 Etn and EC-1 are not expressed in any of the differentiated cells tested,27, 37 and exposure of mouse embryonic stem (ES) cells and embryonic carcinoma (EC) cells to retinoic acid represses expression of EC-1 RNA and induces differentiation.30 In another study ES cells that were incapable of producing miRNA because of a mutated dicer-1 gene remained viable but were severely defective in differentiation.38 The loss of miRNA, which inactivate retroelement RNA,21 could enable ES cells to maintain a high level of retroelement RNA, inhibiting the developmental switch from embryonic cell proliferation to differentiation.
These studies indicate that retroelement RNA, including TSP-binding RNA, are expressed in proliferating embryonic cells and repressed in cells that have stopped proliferating, consistent with a role of the RNA in promoting cell proliferation during embryogenesis.
The first step in the steroidogenic pathway is the conversion of cholesterol to pregnenolone by the enzyme P450scc. Transcription of the P450scc gene is repressed by PSF6, 7, 15 and activated by the binding of a VL30 RNA to PSF.6, 7, 16 Steroidogenesis is induced by pituitary hormones that concomitantly induce synthesis of VL30 RNA in mouse steroidogenic cells,39 suggesting that the primary response of the cells is synthesis of VL30 RNA, which in turn activates transcription of P450scc by binding to PSF. This mechanism enables steroidogenic cells to respond rapidly and reversibly to a hormonal signal. Another relevant study showed that transfection of normal adrenocortical cells with a rasK oncogene derived from a human adrenocortical tumor strongly induced transcription of P450scc and production of cortisol.40 Transcription of P450scc and steroid synthesis also is induced in mouse adrenal cells by VL30 RNA.6 The capacity of a ras oncogene to induce steroidogenesis and also to activate expression of VL30 RNA 32 suggests that the function of a ras oncogene in steroidogenesis involves activating expression of TSP-binding RNA such as VL30 RNA, similar to the function of a ras oncogene in oncogenesis.34