CARMing down the SINEs of anarchy: two paths to freedom from paraspeckle detention

To serve as templates for protein synthesis, messenger RNAs (mRNAs) must be exported from the nucleus in which they are synthesized to the cytoplasm, where their translation occurs. In many human cell types, however, a subset of mRNAs is retained within nuclei and therefore is only inefficiently translated. One feature common to a number of these retained mRNAs is a 3′ untranslated region (UTR) inverted repeat of Alu elements (IRAlus) that forms a hairpin structure containing an imperfectly base-paired stem as long as 300 nucleotides. Alu elements are primate-specific short interspersed elements (SINEs). The degree to which different 3′ UTR IRAlus-containing mRNAs (IRAlus mRNAs) are retained within nuclei varies depending on the particular mRNA and the cell type in which it is expressed. In addition to the widely used process of alternative RNA 3′ end formation that can determine whether a transcript contains a 3′ UTR IRAlus, regulatory mechanisms responsible for variability in the nuclear retention of IRAlus mRNAs exist but are incompletely understood. In the March 15, 2015, issue of Genes & Development, Hu et al. (2015) revealed that the nuclear export of IRAlus mRNAs that can enter nuclear paraspeckles is regulated by a multifunctional arginine methylase.

Paraspeckles are subnuclear sites of IRAlus mRNA retention (Chen et al. 2008). These nonmembrane-delimited bodies are built around the long noncoding RNA NEAT1 (which is essential for paraspeckle integrity) and contain multiple RNA-binding proteins that include non-POU domain-containing octamer binding (p54^nrb or NonO), p54^nrb-associated splicing factor proline/glutamine-rich (PSF or SFPQ), and paraspeckle component 1 (PSPC1) (Fox et al. 2002; Prasanth et al. 2005). In search of new substrates for the coactivator-associated arginine methyltransferase 1 (CARM1; also called PRMT4), Hu et al. (2015) used mass spectrometry and identified p54^nrb in acid-extracted histones from the nuclei of CARM1^−/− mouse embryonic fibroblasts as a substrate for in vitro methylation by Escherichia coli-produced GST-tagged CARM1. Using protein immunostaining and RNA in situ hybridization, p54^nrb and CARM1 were subsequently shown to colocalize within HeLa cell nuclei largely to NEAT1-defined paraspeckles. CARM1-mediated methylation within and near the coiled-coil domain of p54^nrb reduced p54^nrb binding to dsRNAs in vitro and to 3′ UTR IRAlus reporter and cellular mRNAs in HeLa cells. Additionally, replacing endogenous CARM1 with a catalytically inactive variant or replacing endogenous p54^nrb with a variant in which three sites of CARM1-mediated methylation were mutated from arginine to lysine enhanced the retention of 3′ UTR IRAlus reporter and cellular mRNAs within nuclei.

Remarkably, a second path by which CARM1 mediates the retention of reporter and cellular IRAlus mRNAs in paraspeckles was uncovered when the investigators pursued evidence from others that CARM1 associates with different classes of transcription factors. Assays of the abundance of NEAT1 in CARM1 knockdown HeLa cells revealed abnormally more NEAT1 and paraspeckles, and chromatin immunoprecipitations demonstrated that CARM1 is enriched at the NEAT1 promoter. Promoter-bound CARM1 down-regulated NEAT1 gene transcription so as to reduce the number of paraspeckles and thereby promote IRAlus mRNA export to the cytoplasm. Thus, CARM1 “frees” IRAlus mRNAs from paraspeckle “detention” via two paths: methylating p54^nrb and down-regulating NEAT1 gene transcription (Fig. 1). Stimulating NEAT1 synthesis and paraspeckle formation by exposing HeLa cells to poly(I:C), which binds to toll-like receptor 3 and induces an antiviral response (Imamura et al. 2014), was accompanied by reduced CARM1 binding to the NEAT1 promoter. In addition, poly(I:C) treatment reduced CARM1 localization to paraspeckles so as to decrease p54^nrb methylation. As a consequence, poly(I:C) enhanced the nuclear retention of IRAlus mRNA. The mechanism by which poly(I:C) influences cellular CARM1 localization and function remains to be studied.

Data presented by Hu et al. (2015) raise many interesting questions. As examples, what is the structural explanation for the finding that CARM1-mediated methylation of p54^nrb inhibits p54^nrb binding to dsRNA, and how does CARM1 down-regulate NEAT1 synthesis? Another conundrum is why different transcripts with seemingly similar 3′ UTR IRAlus manifest distinctly different subcellular localizations whereby some are efficiently retained in nuclear paraspeckles while others are largely cytoplasmic and either translationally active or translationally silent (Chen et al. 2008; Hundley and Bass 2010; Capshew et al. 2012; Fitzpatrick and Huang 2012; Elbarbary et al. 2013). Competition among the various dsRNA-binding proteins that differentially affect the metabolism of different IRAlus mRNAs undoubtedly plays a role in regulating IRAlus mRNAs on top of which might be layered yet to be discovered post-translational modifications that alter the affinities of dsRNA-binding proteins for IRAlus. For example, it is already known that IRAlus are major binding sites for the dsRNA-binding protein Staufen 1 (STAU1) (Elbarbary et al. 2013; Ricci et al. 2014). When STAU1 outcompetes p54^nrb binding to one class of 3′ UTR IRAlus, then nuclear retention of the affected IRAlus mRNAs is inhibited, and their export to the cytoplasmic is enhanced (Elbarbary et al. 2013). In the cytoplasm, 3′ UTR IRAlus have the potential to bind and activate another dsRNA-binding protein, the translational inhibitor protein kinase R (PKR), making 3′ UTR IRAlus negative regulators of mRNA translation both in cis and in trans (Elbarbary et al. 2013; Kim et al. 2014). STAU1 also competes with PKR for binding to 3′ UTR IRAlus so as to alleviate translational repression (Elbarbary et al. 2013). The segregation of nuclear-retained IRAlus mRNAs away from cytoplasmic PKR during interphase complements the binding of STAU1 to preclude PKR binding to 3′ UTR IRAlus that would otherwise inhibit cellular translation (Elbarbary et al. 2013; Kim et al. 2014). In contrast, dissolution of the nuclear membrane during mitosis removes the boundaries between nuclear-retained IRAlus mRNAs and cytoplasmic PKR, resulting in PKR activation that is necessary for regulated mitosis (Kim et al. 2014).

In summary, genes harboring the potential to encode one or more mRNA isoforms that contain a 3′ UTR IRAlus can be regulated by specialized mechanisms that depend on the efficiencies with which their product mRNAs are retained within the nucleus versus exported to the cytoplasm with or without the concomitant possibility of their translation. These efficiencies, which vary during mitosis and undoubtedly in other cellular states as a means of maintaining cellular homeostasis, are dictated by the potential binding of multiple proteins to 3′ UTR IRAlus within both the nucleus and the cytoplasm. Binding is likely to be regulated by post-translational modifications beyond the CARM1-mediated methylation of p54^nrb described by Hu et al. (2015). Also, while IRAlus are known to be major sites of A-to-I editing and while A-to-I editing does not appear to regulate p54^nrb or STAU1 binding to those IRAlus studied (Elbarbary et al. 2013), editing may influence the binding of other IRAlus factors. It is clear that there is still much to learn about how IRAlus influence mRNA metabolism.