New means to an end: mRNA export activity impacts alternative polyadenylation

ABSTRACT

Gene expression involves multiple co- and post-transcriptional processes that have been increasingly found intertwined. A recent work by our groups (Chen et al. Mol Cell, 2019) indicates that expression of alternative polyadenylation isoforms in mammalian cells can be controlled by nuclear export activities. This regulation has distinct impacts on genes having different sizes and nucleotide contents, and involves RNA polymerase II distribution toward the 3ʹ end of genes. This work raises a number of intriguing questions concerning how 3ʹ end processing and nuclear export are integrated and how their regulation feeds back to transcription.

Gene expression in eukaryotes encompasses a myriad of co- and post-transcriptional events that are elaborately interconnected for streamlined mRNA biogenesis and quality control [1,2]. A recent study by our groups [3] brings into focus the connection between 3ʹ end processing and nuclear export, and the interplays between these steps and RNA polymerase II (Pol II) transcription.

The 3ʹ end processing of almost all protein-coding genes in eukaryotes involves an endonucleolytic cleavage of the nascent RNA and synthesis of the poly(A) tail [4]. These two coupled reactions are carried out by the cleavage and polyadenylation (CPA) complex, including about 20 core factors [4], at the polyadenylation site (PAS). Most genes display alternative cleavage and polyadenylation (APA) at multiple PASs [5,6]. Some APA sites are in introns, leading to isoforms with different coding sequences; most APA sites are in the last exon, affecting the 3ʹUTR length of expressed mRNAs [7]. Accumulating studies have shown that APA is dynamically regulated, often in a global manner, in various cell conditions, such as cell proliferation, differentiation, neuronal activation, stress, etc [8].

Nuclear export of mRNA, a prerequisite for its cytoplasmic metabolism, is mediated by the evolutionarily conserved transport receptor NXF1-NXT1, a heterodimer complex that interacts with nucleoporins along the central channel of the nuclear pore for mRNA translocation from nucleus to cytoplasm. Interestingly, the intrinsic RNA binding activity of NXF1-NXT1 is quite weak [9], a property seemingly unfit for its role in mRNA export. However, adaptor proteins, with which NXF-NXT1 interacts, confer RNA binding affinities and specificities [10,11] (Figure 1). A growing number of the adaptor proteins have been uncovered in recent years, most, if not all, of which interact with substrate mRNAs at processing steps. For example, ALYREF, part of the highly conserved transcription-export (TREX) complex [12], is recruited to nascent RNAs by the nuclear cap-binding complex (CBC), poly(A)-binding protein (PABPN1), and the exon-junction complex (EJC) [13–15]; Serine and arginine domain-containing proteins (SR proteins) SRSF3 and SRSF7 facilitate nuclear export of select mRNAs through splicing and/or CPA [11,16]; CFI-68, part of the cleavage factor I subcomplex of CPA machinery and an SR-like protein, also physically interacts with NXF1 [17].

Figure 1.

Schematic showing connections between nuclear export factors and 3ʹ end processing and transcription. NXF1-NXT1 mediates mRNA nuclear export through various adaptors binding to newly made RNA. The UGUA motif are enriched for long 3ʹUTR isoforms, owing to its preferential presence before the distal PASs. CFI-68 binds to UGUA to promote distal PAS usage, and enhances long isoform nuclear export through its function as an adaptor for NXF1-NXT1. By contrast, short 3ʹUTR isoforms are more efficiently exported, even without the help of CFI-68. NXF1 also appears to regulate Pol II elongation of select genes (large size and high AT content) at the 3ʹ end, which impacts PAS choice.

The role of CFI-68 in mRNA export was extensively studied in our work [3]. We found that transcript length and 3ʹ UTR size are major features of the transcripts whose nuclear export is affected after CFI-68 knockdown (KD) [3]. As such, CFI-68 preferentially facilitates nuclear export of long 3ʹ UTR isoforms (Figure 1), a role consistent with its interaction with NXF1. Importantly, this target-specific role of CFI-68 is elegantly aligned with the enrichment of CFI-68 binding motif UGUA near distal PASs (Figure 1), a feature also important for distal PAS usage [18,19]. Therefore, CFI-68 and UGUA motifs play dual roles in enhancing both CPA and nuclear export for long 3ʹ UTR mRNAs.

Surprisingly, NXF1 KD results in global increase of short 3ʹ UTR isoform expression and concomitant decrease of long 3ʹUTR isoform expression, similar to CFI-68 KD [18–20]. The APA isoform abundance changes is likely due to PAS usage regulation rather than differential isoform degradation: genes with shortened 3ʹ UTRs in NXF1 KD cells showed slight upregulation, refuting the possibility that 3ʹUTR shortening is caused by long isoform degradation; detailed analysis of two genes (SIAH2 and NFKB1) did not show mRNA decay difference between NXF1 KD and control cells.

So, does NXF1 directly regulate CPA activity? An in vitro pre-mRNA cleavage assay using nuclear extract of NXF1 KD cells did not indicate CPA activity changes, as in that of CFI-68 KD cells. While we cannot completely rule out the possibility that NXF1 KD may still inhibit CPA in vivo, the fact that transcriptional readthrough, often a consequence of defective CPA [21], is not observed with the chromatin immunoprecipitation and deep sequencing (ChIP-seq) does not lend support to this model. Notably, while NXF1 KD and CFI-68 KD cells showed similar global trends in 3ʹUTR shortening, there are many APA events unique to each condition, indicating distinct mechanisms.

So, what is causing PAS selection changes in NXF1 KD cells? By ChIP-seq analysis of Pol II distribution and sequencing of chromatin-bound nascent RNA, we found that Pol II in NXF1 KD cells tends to pile up along the gene body, especially toward the 3ʹ end. Intriguingly, this trend is most obvious for genes with large sizes and in high AT content regions. Importantly, genes with these features also display the greatest 3ʹ UTR shortening after NXF1 KD. These results thus indicate that limiting nuclear export activity by decreasing NXF1 level may affect Pol II elongation in select genes. This finding is reminiscent of previous observations where loss of transcription elongation factors or expression of slow Pol II mutants shift PAS usage toward proximal ones in yeast and fruit flies [22–24]. In sum, our work posits a more active role of NXF1 in upstream mRNA biogenesis events such as Pol II elongation and 3ʹ end formation than previously thought, highlighting the coordination among Pol II dynamics, 3ʹ end processing and nuclear export.

Outstanding questions

A potential ramification of NXF1-mediated APA regulation is a feedback mechanism by which mRNA export activity registers with 3ʹ UTR size (Figure 1): when nuclear export activity is low, creating an unfavorable environment for long 3ʹUTR isoform export, short 3ʹUTR isoform production is enhanced. Conversely, when nuclear export activity is high, both the biogenesis and export of long 3ʹUTR isoforms are elevated. Our work also raises a few questions for further investigation:

1. While it is evident that NXF1 KD leads to Pol II pileup, especially toward the 3ʹ end, indicative of defects in transcriptional elongation, the detailed mechanisms remain elusive. Unlike the 5ʹ end of genes, Pol II elongation rate at the 3ʹ end region is so far poorly understood. The physical interaction between NXF1 and Pol II seems to suggest a role of NXF1 in Pol II activity. One possibility is that NXF1 mediates regulation of the CTD phosphorylation, such as serine 2 and threonine 4 positions of CTD heptads, which are, respectively, linked to elongation and termination [25]. The select effect on genes with large sizes and high AT contents are of special note. In line with the notion that transcriptional dynamics can be different for genes of different sizes, Pol II elongation rate has been reported to increase in the first 10–15 kb of the gene body [26], and expression of long genes appears to have a greater dependency on the Pol II processivity regulator Spt5 [27]. In addition, genes in high AT content regions have also been found to have a higher elongation rate at the basal level [28]. Therefore, it is possible that Pol II elongation over long genes and in high AT content regions is especially sensitive to NXF1 levels. In this vein, gene size was recently found to be an important feature for APA regulation by PCF11 [29]. Whether APA regulatability of a gene generally registers with Pol II elongation rate needs to be further examined.

2. In keeping with the intertwined connections between 3ʹ end processing and nuclear export, there are other interactions between CPA factors and nuclear export factors. For example, ALYREF interacts with the CPA factor CstF-64 [14], THOC5 co-purifies with CFI-68 [30], and yeast Pcf11 interacts with Yra1, a homolog of ALYREF [31]. Our work in fact also found that KDs of several nuclear exports, such as THOC2 and ALYREF, led to 3ʹ UTR shortening, albeit to a lesser extent as compared to NXF1 KD. And, another TREX member CHTOP was recently shown to influence PAS choice [15]. Conversely, the Proudfoot lab previously reported that phosphorylation of PCF11 by WNK1 regulates nuclear export [32]. Therefore, how other nuclear export factors play roles in 3ʹ end processing and how other CPA/termination factors are involved in nuclear export are interesting directions to explore in the future.

3. In line with a previous study [33], we show that mRNAs with long 3ʹ UTRs are less efficiently exported to the cytoplasm. The mechanistic basis for this, however, is not clear. Neve et al. indicated that some of the nuclear-retained distal PAS isoforms may represent incompletely spliced transcripts, leading to their inefficient export [33]. Whether there is a global connection between 3ʹ UTR-controlled nuclear export and splicing is yet to be established. On this note, we recently found that SF3b, a core splicing factor, balances mRNA export with processing [34]. Therefore, an emerging theme appears to be that the nuclear export potential of a given transcript is determined by various processing steps it experiences. In this vein, given that 3ʹ UTR is devoid of export adaptors deposited through splicing due to lack of intron sequences (evolutionarily selected against because of the nonsense-mediated decay, or NMD, mechanism), it may especially dependent on UGUA motif abundance for efficient nuclear export. Whether there are other motifs playing a similar role needs to be further studied.

4. APA regulation has been reported in different tissue types, cell proliferation and differentiation and stress conditions [8,35]. To what extent NXF1-based APA regulation plays a role in these conditions would be an interesting question to address in the future. Conversely, how this mechanism is connected to cytoplastic metabolisms, such as mRNA stability, translation, and localization, especially in cell types displaying extreme 3ʹ UTR size preferences, such as neurons and immune cells [36,37], await further investigations.

Funding Statement

This work was supported by the National Institutes of Health [GM084089 and GM129069]; National Natural Science Foundation of China [31570822, 91540104, and 31770880]; Strategic Priority Research Program of the Chinese Academy of Sciences [XDB19000000]; National Key R&D Program of China [2017YFA0504400].

Acknowledgments

We thank members of BT and HC labs for helpful discussions. This work was funded by the National Key R&D Program of China [2017YFA0504400], Strategic Priority Research Program of the Chinese Academy of Sciences (XDB19000000), National Natural Science Foundation of China (31570822 and 31770880) to H.C., and NIH (GM084089 and GM129069) to BT.

Disclosure statement

No potential conflict of interest was reported by the authors.