Regulated circRNA Nuclear Export in Neuronal Differentiation

Abstract

Multiple mechanisms have been reported on how circular RNAs (circRNAs) are exported to the cytoplasm. A recent paper by Cao et al. has shown that export of a subset of circRNAs with (A)-rich motifs, including one with a clear function, is regulated during neuronal development via a novel mechanism.

Differentiation-dependent nuclear exit of (A)-rich circRNAs

Circular RNAs (circRNAs) are formed by back-splicing of pre-mRNAs in the nucleus and have been reported to have a variety of functions in both normal and cancer cells [1]. While generally produced at low rates, their exceptional stability owing to resistance to exonucleases results in much higher levels of accumulation in nondividing cells such as mature neurons, suggesting potential functional roles in such cells. Most circRNAs accumulate not in the nucleus but in the cytoplasm. As most circRNAs are generated from pre-mRNAs by the same splicing machinery as mRNAs, does their export to the cytoplasm utilize the same machinery as well? A number of recent studies reveal that this is not necessarily the case. These RNAs differ from mRNAs in that they lack 5’-cap structures and poly(A) tails, and it is not known whether they even harbor exon-junction complexes. The process of mRNA export begins by the cotranscriptional association of various RNA-binding proteins, most notably the TREX complex, which then recruits the export receptor complex NXF1-NXT1 to assemble export-competent mRNPs [2]. Importantly, much mRNA export relies on recognition of both ends of the mRNA - the 5’-cap structure and 3’-polyadenylation. Generation of the poly(A) tail is associated with deposition of the nuclear poly(A)-binding protein PABPN1 and proper polyadenylation is required for efficient mRNA export from the nucleus [3,4]. A common view is that PABPN1-bound nuclear mRNAs are remodeled during or after nuclear export so that PABPN1 is replaced by the cytoplasmic poly(A)-binding protein, PABPC1, to promote robust cap-dependent translation in the cytoplasm [2]. Before now, there has been little functional evidence reported to support a nuclear function for PABPC1 in mammals.

In the past few years a surprising variety of mechanisms have been reported to explain how circRNAs can be exported to the cytoplasm. Specific protein binding motifs can promote export [5]. Components of the TREX complex, DDX39A and DDX39B, were seen to influence circRNA export but export depended on the length of the circRNAs [6]. Further, the m6A reader YTHDC1 was found to help mediate the export of circRNAs that harbor m6A modification [7]. Recent work also identified export pathways that are analogous to protein export, rather than mRNA export. These involve exportin-4 (XPO4) and exportin-2 (XPO2) [8,9]. For example, IGF2BP1 has been shown to bind directly to some circular RNAs and recruit Ran-GTP and XPO2 to mediate export [9].

Cao et al. [10] have approached the circRNA export issue in a new way. These authors examined the subcellular localization of circRNAs during neuronal differentiation and observed a subset of circRNAs that were retained in the nucleus in pluripotent human H9 cells but exported to the cytoplasm in forebrain neurons (FN) differentiated from H9 cells. Importantly, this striking differentiation-dependent subcellular relocation was not seen for mRNAs and long noncoding RNAs (lncRNAs). Inspection of this regulated subset revealed a significant enrichment of (A)-rich motifs in the nuclear circRNAs and these motifs were found to associate with not the nuclear PABPN1, but with PABPC1. In complexes of circRNAs with PABPC1, the authors also found the nuclear basket protein TPR, which is known to be involved in the quality control of RNA export. By microscopy, nuclear circRNA complexes appeared to be tethered to nuclear pores and thus blocked from export. Reduction of levels of either PABPC1 or TPR resulted in the export of the (A)-rich circRNAs without affecting that of their linear counterparts. A likely mechanism for the regulation of (A)-rich circRNA export during differentiation was revealed by data showing that in H9 cells there was a higher nuclear concentration of PABPC1 than in FN. Thus, regulation of nuclear PABPC1 correlated with regulated circRNA export. Functionally, one circRNA, circRTN4 [1,2], which is specifically expressed at high levels in the brain, was nuclear in H9 cells but cytoplasmic in FN. In three independent experiments, knockdown of circRTN4 [1,2] led to shortened neurite outgrowth. This latter result highlighted a novel and critical functional importance of regulated circRNA nuclear export in neuronal development (Figure 1). Finally, Cao et al. asked whether the regulated retention of bulk (A)-rich circRNAs might be related to regulation of mRNA translation rates. In this model, pluripotent cells retain (A)-rich circRNAs in the nucleus to avoid competitive binding of their (A)-rich motifs with cytosolic PABPC1, ensuring the maximum possible protein translation activity. H9 cells are proliferative and highly active in translation while neurons are not. Abundant (A)-rich circRNAs in FN may help to dampen translation in these cells and/or could impact mRNA decay, another process associated with PABPC1. Some evidence supporting this translation model was presented.

Figure 1.

In pluripotent cells, a significant fraction of PABPC1 is in the nucleus, where it binds to (A)-rich circRNAs and subsequently tethers them to TPR at nuclear pores, preventing export. In forebrain neurons, nuclear PABPC1 levels are lower, allowing more efficient export of (A)-rich circRNAs. One of these, circRTN4 [1,2], promotes neurite outgrowth.

The ramifications of this study are intriguing and many questions remain. It is not known whether nuclear (A)-rich circRNA accumulation is a phenomenon specific to human pluripotent cells and whether the potential role of (A)-rich circRNAs in regulation of translation also operates in non-neuronal cells. Mechanistically, how circRTN4 exerts its effect on neurite outgrowth is unclear. Do circRNAs other than circRTN4 play roles in differentiation of pluripotent cells to other tissues or cell types? If so, are there specific sequence elements or motifs in some of these circRNAs that contribute to function in neuronal development? Finally, this study clearly illustrates that much more needs to be learned about the functional significance of PABPC1-TPR interactions and the broader issue of nuclear PABPC1 function. While the mechanism of nuclear PABPC1 accumulation remains to be investigated, it is not clear whether nuclear PABPC1 is associated with gene regulation in more ways than has heretofore been appreciated, and possibly in more cells and tissues.