Portal de Periódicos da CAPES

Regulating the Regulators: Posttranslational Modifications of RNA Silencing Factors

2009; Cell Press; Volume: 139; Issue: 1 Linguagem: Inglês

10.1016/j.cell.2009.09.013

1097-4172

Inha Heo, V. Narry Kim,

MicroRNA in disease regulation

Every regulator should be regulated, and this holds true for small RNAs and their associated proteins. Knowledge has begun to emerge of the various mechanisms that impose specificity on the expression and function of RNA silencing factors. Recent papers, including one in this issue of Cell (Paroo et al., 2009Paroo Z. Ye X. Chen S. Liu Q. Cell. 2009; (this issue)PubMed Google Scholar), now reveal the posttranslational modifications that take part in the regulation of the core RNA silencing factors, Ago, Piwi, and TRBP. Every regulator should be regulated, and this holds true for small RNAs and their associated proteins. Knowledge has begun to emerge of the various mechanisms that impose specificity on the expression and function of RNA silencing factors. Recent papers, including one in this issue of Cell (Paroo et al., 2009Paroo Z. Ye X. Chen S. Liu Q. Cell. 2009; (this issue)PubMed Google Scholar), now reveal the posttranslational modifications that take part in the regulation of the core RNA silencing factors, Ago, Piwi, and TRBP. Small RNAs of 20–30 nt direct gene regulation through a wide range of mechanisms: from DNA methylation and heterochromatin formation to mRNA destabilization and translational control. Through such extensive patrolling over the genome and transcriptome, small RNAs are involved in numerous biological processes, including developmental timing, cell differentiation, cell proliferation, cell death, metabolic control, immunity, and transposon silencing (Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar). A defining feature of small RNAs is their association with Argonaute family proteins that act as the effectors in RNA silencing (Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar). Argonaute family proteins are classified into two clades: Ago and Piwi subfamilies. Based on the types of Argonaute proteins that they are associated with and their biogenesis pathway, animal small RNAs are grouped into three classes: microRNAs (miRNAs), endogenous small-interfering RNAs (endo-siRNAs), and piwi-interacting RNAs (piRNAs). MicroRNAs are generated from local hairpin structures by the action of two RNase III type proteins, Drosha and Dicer. Mature miRNAs of ∼22 nt are then incorporated into Ago subfamily proteins and act as posttranscriptional regulators by targeting mRNAs. Piwi-interacting RNAs (24–30 nt in length), on the other hand, are bound to Piwi subfamily proteins, are highly abundant in germline cells and function in transposon silencing through heterochromatin formation or RNA destabilization. Endo-siRNAs (∼21 nt) have been studied mostly in invertebrates, but they have also been found in mouse oocytes and embryonic stem (ES) cells. Endo-siRNAs are similar to miRNAs in their association with Ago subfamily members and their mode of action as posttranscriptional regulators. They differ from miRNAs, however, in that they are produced from long dsRNAs by Dicer, and in that endo-siRNAs suppress transposable elements whereas miRNAs control protein-coding genes. Recent reports have begun to unveil various ways of regulating silencing factors. In this Minireview, we will focus on the posttranslational modifications of protein factors involved in small RNA pathways. Canonical animal miRNAs are generated via two-step processing (Figure 1A). The nuclear RNase III Drosha and its cofactor DGCR8/Pasha cleave the primary transcript of an miRNA gene (pri-miRNA) into a hairpin-structured precursor (pre-miRNA) (Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar). The pre-miRNA is then exported to the cytoplasm and gets further processed into a mature miRNA of ∼22 nt by another RNase III, Dicer. Following Dicer processing, one strand of the RNA duplex is loaded onto the RNA-induced silencing complex (RISC) that contains the Argonaute protein as the core component. Dicer forms a processing complex with a partner protein containing dsRNA-binding domains (dsRBDs) (Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar). In Drosophila, Dicer-1 requires Loquacious (Loqs, also known as R3D1) for pre-miRNA processing while Dicer-2 binds to R2D2 to form a stable complex to generate siRNAs. Human Dicer interacts with two closely related proteins, TRBP (HIV-1 TAR RNA-binding protein) and PACT, both of which contain three dsRBDs. Although neither TRBP nor PACT is required for the Dicer processing reaction itself, they seem to stabilize Dicer and contribute to RISC formation. Paroo et al. report in this issue of Cell that TRBP is positively regulated by the mitogen-activated protein kinase (MAPK) pathway (Figure 1B). The study began with the observation that in western blotting the TRBP protein appeared as multiple bands that disappeared upon phosphatase treatment. By mass spectrometry and mutagenesis, the authors discovered that the TRBP protein is phosphorylated at four serine residues (142, 152, 283, and 286). Computational algorithms predicted that TRBP may be a substrate for one of the mitogen-activated protein kinases (MAPKs), Erk, which is a major determinant in cellular proliferation, differentiation, and survival (Chen et al., 2001Chen Z. Gibson T.B. Robinson F. Silvestro L. Pearson G. Xu B. Wright A. Vanderbilt C. Cobb M.H. Chem. Rev. 2001; 101: 2449-2476Crossref PubMed Scopus (738) Google Scholar). Indeed, TRBP interacted with phosphorylated Erk in vivo and was phosphorylated under conditions where Erk was activated. In order to reveal the physiological function of TRBP phosphorylation, the authors generated isogenic cell lines that express wild-type, phospho-mutant, and phopho-mimic TRBPs. Notably, cell line expressing phospho-mimic TRBP showed elevated levels of TRBP, leading to an increase in the Dicer protein, presumably through protein stabilization. As a consequence of the increase in TRBP and Dicer, mature miRNA levels increased. Interestingly, most miRNAs (particularly those that facilitate cell proliferation and growth) were upregulated in the cells expressing phospho-mimic TRBP, whereas the let-7 family miRNAs, the known suppressor of cell growth, were downregulated. Consistently, a specific inhibitor of the Erk pathway, U0126, decreased pro-growth miRNAs and increased anti-growth let-7 miRNAs in several cancer cell lines. Cell growth and survival rate were found to be increased in a TRBP phosphorylation-dependent manner upon Erk activation. The reason for the suppressive effect on let-7 is unclear but it may be that Erk signaling represses let-7 indirectly through regulators such as Lin28 (Dangi-Garimella et al., 2009Dangi-Garimella S. Yun J. Eves E.M. Newman M. Erkeland S.J. Hammond S.M. Minn A.J. Rosner M.R. EMBO J. 2009; 28: 347-358Crossref PubMed Scopus (296) Google Scholar). Thus, the function of the Erk signaling pathway is achieved in part through the control of the miRNA biogenesis pathway. This study for the first time establishes protein modification as the molecular link between an important cell signaling pathway and the miRNA machinery. The first demonstration of a posttranslational modification of RNA silencing factors appeared last year. From a proteomic analysis of human Ago proteins, Qi et al. found that Ago2 interacts with the subunits of the type I collagen prolyl-4-hydroxylase (C-P4H(I)) (Qi et al., 2008Qi H.H. Ongusaha P.P. Myllyharju J. Cheng D. Pakkanen O. Shi Y. Lee S.W. Peng J. Nature. 2008; 455: 421-424Crossref PubMed Scopus (166) Google Scholar). Mass spectrometry revealed that Ago2 is hydroxylated at proline 700. Importantly, hydroxylation enhanced the stability of the Ago2 protein and knockdown of C-P4H(I) reduced siRNA-induced silencing activity (Figure 1C). Putative hydroxylation sites were found in other human Ago proteins (Ago1, Ago3, and Ago4) and in mouse and fly Ago proteins, but it remains unknown whether Ago hydroxylation is a conserved process. Apart from stabilizing Ago2, hydroxylation enhanced the localization of Ago2 to cytoplasmic foci called P bodies (Qi et al., 2008Qi H.H. Ongusaha P.P. Myllyharju J. Cheng D. Pakkanen O. Shi Y. Lee S.W. Peng J. Nature. 2008; 455: 421-424Crossref PubMed Scopus (166) Google Scholar). Of note, Ago2 localization seems to be influenced by phosphorylation as well. Zeng et al., 2008Zeng Y. Sankala H. Zhang X. Graves P.R. Biochem. J. 2008; 413: 429-436Crossref PubMed Scopus (158) Google Scholar observed that human Ago2 becomes phosphorylated at serine 387 in response to the activation of the p38 MAPK pathway. Ago2 phosphorylation does not alter the level of Ago2 protein, but it facilitates the localization of Ago2 to P bodies. P bodies are concentrated with RNAs and proteins involved in mRNA decay and translational regulation (Eulalio et al., 2007Eulalio A. Behm-Ansmant I. Izaurralde E. Nat. Rev. Mol. Cell Biol. 2007; 8: 9-22Crossref PubMed Scopus (716) Google Scholar). There is some evidence that miRNA- and siRNA-mediated silencing occurs in P bodies. However, disruption of the P body does not affect silencing, implying that neither P bodies themselves nor Ago2 localization to the P bodies is essential for the RNAi pathway (Eulalio et al., 2007Eulalio A. Behm-Ansmant I. Izaurralde E. Nat. Rev. Mol. Cell Biol. 2007; 8: 9-22Crossref PubMed Scopus (716) Google Scholar). Therefore, the physiological significance of P body localization induced by Ago modification remains unclear. The second group of Argonaute proteins is the Piwi subfamily that takes part in the biogenesis and function of piRNAs in animal gonads (Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar). Mice express three Piwi proteins: Mili, Miwi, and Miwi2. Drosophila also has three Piwi proteins: Piwi, Aub, and Ago3. Piwi-interacting RNAs are processed from single-stranded RNA precursors that are transcribed from transposons or large piRNA clusters. In mice, piRNAs found in pre- and neonatal stages (pre-pachytene piRNAs) require Mili and Miwi2 and are derived largely from repeat elements. Pre-pachytene piRNAs contribute to transposon silencing through DNA methylation and possibly RNA cleavage. Piwi-interacting RNAs in adult testes (from the pachytene stage of male germ cells) are dependent on Mili and Miwi. Pachytene piRNAs are derived mainly from intergenic loci in the genome and from piRNA clusters. The function of pachytene piRNAs remains unknown. Piwi-interacting RNA biogenesis involves two distinct mechanisms (known as primary and secondary) (Figure 1D) (Aravin et al., 2008Aravin A.A. Sachidanandam R. Bourc'his D. Schaefer C. Pezic D. Toth K.F. Bestor T. Hannon G.J. Mol. Cell. 2008; 31: 785-799Abstract Full Text Full Text PDF PubMed Scopus (766) Google Scholar, Kim et al., 2009Kim V.N. Han J. Siomi M.C. Nat. Rev. Mol. Cell Biol. 2009; 10: 126-139Crossref PubMed Scopus (2404) Google Scholar, Siomi and Kuramochi-Miyagawa, 2009Siomi M.C. Kuramochi-Miyagawa S. Curr. Opin. Cell Biol. 2009; 21: 426-434Crossref PubMed Scopus (28) Google Scholar). In primary processing, the piRNA precursor is cleaved by an unknown enzyme(s) in the cytoplasm and the product is loaded on to Mili and Miwi. In secondary processing (known as the ping-pong cycle), Mili and Miwi2 use their own endonucleolytic activity to generate and amplify piRNAs. Pachytene piRNAs are thought to be generated through primary processing. Pre-pachytene piRNAs are produced through primary processing and amplified via the ping-pong cycle. The first indication of Piwi modification came serendipitously when Kirino et al. realized that the widely used Y12 monoclonal antibody recognizes Mili and Miwi proteins (Kirino et al., 2009Kirino Y. Kim N. de Planell-Saguer M. Khandros E. Chiorean S. Klein P.S. Rigoutsos I. Jongens T.A. Mourelatos Z. Nat. Cell Biol. 2009; 11: 652-658Crossref PubMed Scopus (174) Google Scholar). Knowing that the epitope of Y12 consists of symmetrically dimethylated arginines (sDMAs), the authors searched for sDMA motifs (typically Gly-Arg-Gly) in Piwi proteins and found that methylation occurs at conserved sDMA motifs in the N terminus of Aub. Most Piwi proteins in mouse, Drosophila, and Xenopus contain such sDMA motifs. The precise methylation sites in Mili and Miwi were later mapped by other groups by mass spectrometry (Reuter et al., 2009Reuter M. Chuma S. Tanaka T. Franz T. Stark A. Pillai R.S. Nat. Struct. Mol. Biol. 2009; 16: 639-646Crossref PubMed Scopus (202) Google Scholar, Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). The Drosophila mutant deficient in the protein methyltransferase 5 gene (dPRMT5, also called capsuleenn [csul] or dart5) was known to phenocopy the aub null mutant, implicating a link between these two genes. Kirino et al. showed that arginine methylation of Piwi proteins was indeed abrogated in dPRMT5-deficient ovary (Kirino et al., 2009Kirino Y. Kim N. de Planell-Saguer M. Khandros E. Chiorean S. Klein P.S. Rigoutsos I. Jongens T.A. Mourelatos Z. Nat. Cell Biol. 2009; 11: 652-658Crossref PubMed Scopus (174) Google Scholar). The protein levels of Aub, Ago3, and Piwi were reduced without changes at the mRNA levels, suggesting that arginine methylation stabilizes the Piwi proteins. The levels of piRNAs also decreased and transposons were derepressed (Figure 1E). The Drosophila embryo contains a specialized region known as the pole plasm at its posterior end. The pole plasm is important for the specification of primordial germ cells (PGCs) (Strome and Lehmann, 2007Strome S. Lehmann R. Science. 2007; 316: 392-393Crossref PubMed Scopus (145) Google Scholar). In dPRMT5-deficient flies, the level of Aub in the pole plasm is strongly reduced, suggesting that Aub methylation may have an important role in pole plasm assembly and PGC formation. Several questions then followed: what recognizes the sDMA in Piwi proteins and what are the consequences of sDMA recognition? The answers soon emerged when several groups independently carried out purification and proteomic analyses on Piwi complexes (Reuter et al., 2009Reuter M. Chuma S. Tanaka T. Franz T. Stark A. Pillai R.S. Nat. Struct. Mol. Biol. 2009; 16: 639-646Crossref PubMed Scopus (202) Google Scholar, Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). From the analyses of testes from transgenic mice expressing epitope-tagged Piwi proteins, Vagin et al. found that Piwi complexes contain not only PRMT5 and its cofactor WDR77/MEP50/WD45 but also several Tudor domain-containing proteins (TDRDs) (Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). A similar approach, using monoclonal antibody against Mili, identified TDRD1 as the interactor (Reuter et al., 2009Reuter M. Chuma S. Tanaka T. Franz T. Stark A. Pillai R.S. Nat. Struct. Mol. Biol. 2009; 16: 639-646Crossref PubMed Scopus (202) Google Scholar, Wang et al., 2009Wang J. Saxe J.P. Tanaka T. Chuma S. Lin H. Curr. Biol. 2009; 19: 640-644Abstract Full Text Full Text PDF PubMed Scopus (130) Google Scholar). Tudor domains are known to recognize methylated arginines, and TDRDs are essential for germ cell development in both mouse and fly (Thomson and Lasko, 2005Thomson T. Lasko P. Cell Res. 2005; 15: 281-291Crossref PubMed Scopus (66) Google Scholar, Hosokawa et al., 2007Hosokawa M. Shoji M. Kitamura K. Tanaka T. Noce T. Chuma S. Nakatsuji N. Dev. Biol. 2007; 301: 38-52Crossref PubMed Scopus (117) Google Scholar). TDRD1 interacts strongly with Mili, but it also binds to Miwi2 and Miwi. Deep sequencing of piRNAs from embryonic testes revealed that transposon-derived pre-pachytene piRNAs were reduced in the Tdrd1-deficient mice (Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar), suggesting that TDRD1 contributes to pre-pachytene piRNA biogenesis. Repeat elements, such as LINE, were derepressed in the Tdrd1 mutant testes (Reuter et al., 2009Reuter M. Chuma S. Tanaka T. Franz T. Stark A. Pillai R.S. Nat. Struct. Mol. Biol. 2009; 16: 639-646Crossref PubMed Scopus (202) Google Scholar, Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). TDRD1 also influences the localization of Miwi2 to a cytoplasmic structure called nuage that is important for germ cell development. In the Tdrd1 mutant, Miwi2 delocalizes from the nuage and disperses to the cytoplasm. Consistent with this, the transposon-derived piRNAs that are associated with Miwi2 were more dramatically reduced than other piRNAs (Figure 1F) (Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). Miwi, on the other hand, interacts with TDRD6 in postnatal testes. In round spermatids, Miwi mainly localizes to the chromatoid body, a structure related to nuage, and functions in the pachytene piRNA pathway along with Mili. In the Tdrd6-deficient mutant, Miwi loses its chromatoid body localization and disperses to the cytoplasm. However, the piRNA population is not altered, and transposons seem to be unaffected (Vasileva et al., 2009Vasileva A. Tiedau D. Firooznia A. Muller-Reichert T. Jessberger R. Curr. Biol. 2009; 19: 630-639Abstract Full Text Full Text PDF PubMed Scopus (133) Google Scholar). These results suggest that chromatoid body formation may be independent of piRNAs. Some miRNAs were found to accumulate in the Tdrd6 mutant (Vasileva et al., 2009Vasileva A. Tiedau D. Firooznia A. Muller-Reichert T. Jessberger R. Curr. Biol. 2009; 19: 630-639Abstract Full Text Full Text PDF PubMed Scopus (133) Google Scholar), but this could be explained by the timing of development in the mutant (Vagin et al., 2009Vagin V.V. Wohlschlegel J. Qu J. Jonsson Z. Huang X. Chuma S. Girard A. Sachidanandam R. Hannon G.J. Aravin A.A. Genes Dev. 2009; 23: 1749-1762Crossref PubMed Scopus (228) Google Scholar). Thus, the significance of the interaction between Miwi and TDRD6 is currently unclear. The sDMA marks on Piwi are read and interpreted by different Tudor domain proteins. Multiple Piwi proteins seem to interact with multiple TDRDs (at least six TDRDs were precipitated with mouse Piwi proteins). However, there seems to be a certain degree of specificity in the interactions between Piwi proteins and TDRDs. It will be necessary to identify the physiological partners of each Piwi protein. It would also be important to uncover the biochemical functions of TDRD in order to fully understand the role of arginine methylation in the piRNA pathway. An emerging theme in a recent string of literature is that small RNA pathways are under intensive regulation. Multiple layers of regulation ensure tight control of silencing, and dysregulation of small RNAs may lead to human diseases such as cancer, myopathy, and neurodegeneration. Posttranslational modification has been established as one of the key regulatory layers. It will be interesting to further identify new types of modifications. It will also be important to reveal the upstream signaling pathways that affect small RNAs. Developmental and environmental cues need to be accurately transduced to RNA silencing machinery to maintain the normal cellular function. Many miRNAs appear to be controlled posttranscriptionally through unknown mechanisms. Such miRNAs may be controlled by cell signaling pathways, possibly via posttranslational modification of the miRNA machinery. Whether Piwi methylation by PRMT5 is controlled by external signals also awaits further investigation. It will be an exciting quest to uncover the missing links between cell signaling and RNA silencing. We thank D. Jee, C. Joo, and Y.-K. Kim for helpful discussions. This work was supported by the Creative Research Initiatives Program (20090063603) through the National Research Foundation of Korea (NRF) and the BK21 Research Fellowship (I.H.) from the Ministry of Education, Science and Technology of Korea.