Re-editing the paradigm of Cytidine (C) to Uridine (U) RNA editing

Abstract

Cytidine (C) to Uridine (U) RNA editing is a post-trancriptional modification that until recently was known to only affect Apolipoprotein b (Apob) RNA and minimally require 2 components of the C to U editosome, the deaminase APOBEC1 and the RNA-binding protein A1CF. Our latest work has identified a novel RNA-binding protein, RBM47, as a core component of the editosome, which can substitute A1CF for the editing of ApoB mRNA. In addition, new RNA species that are subjected to C to U editing have been identified. Here, we highlight these recent discoveries and discuss how they change our view of the composition of the C to U editing machinery and expand our knowledge of the functional attributes of C to U RNA editing.

Cytidine (C) to Uridine (U) RNA editing was first reported for Apolipoprotein b (Apob) in the late 1980s.^1,2 There are 2 major isoforms of APOLIPOPROTEIN B proteins: APOB100 that has a molecular weight of about 500 kDa and is produced in the liver, and a shorter version, APOB48, synthesized in the small intestine. After ruling out the possibilities that these protein isoforms were produced by 2 different genes or that APOB48 was a cleavage product of APOB100, sequence analysis of 14121-nucleotide-long Apob messenger (m)RNA revealed that APOB48 was in fact the product of a then new process of post-transcriptional modification. The C at position 6666 of Apob mRNA is changed to U, creating a premature UAA stop codon such that the edited transcript codes for APOB48 instead of APOB100 in the small intestine.^1,2 The deaminase APOBEC1 was subsequently discovered and was shown to be responsible for the deamination of C6666 into U.^3-5 Consistent with this role, loss of Apobec1 function in mutant mice abolishes C to U editing of Apob mRNA.^6-8 APOBEC1 is the catalytic subunit of a multi-protein complex referred as the C to U editosome.^9,10 While APOBEC1 can directly bind RNA, in vitro experiments suggest that it is not sufficient to elicit editing. Purified recombinant APOBEC1 is unable to edit a synthetic unedited Apob RNA by itself and requires a cofactor for this process.^10-12 Until recently, the RNA binding protein (RBP) A1CF (also known as ACF) was the only factor known to partner with APOBEC1 to mediate C to U RNA editing,^10-16 leading to the concept of the minimal core components of the C to U editosome: APOBEC1 and A1CF for editing of Apob RNA. Our recent work has identified a novel RBP, RBM47, that can substitute A1CF in vitro and is essential for C to U RNA editing in vivo.¹⁷ In addition, new RNA targets for APOBEC1 editing have been identified through sequencing analysis of the transcriptome of Apobec1-null mutant mice.^18,19 Here, we will discuss how these discoveries call for a revised view of the core components of the editing machinery and open new avenues to study the role of C to U RNA editing.

RBM47, a novel requisite cofactor for C to U RNA editing

Rbm47 was identified in a screen as one of the genes that are differentially expressed in the foregut endoderm of embryonic day (E) 8.5 mouse embryo.²⁰ In addition to the embryonic foregut endoderm, Rbm47 is expressed in adult organs such as the liver, the pancreas, the lung and the small intestine.¹⁷ RBM47 protein is conserved between human and mouse and has 3 RNA recognition motifs (RRM). RBM47 was shown to bind RNA in the human epithelial colorectal adenocarcinoma Caco-2 cell line.¹⁷ Analysis of the ENSEMBL database²¹ revealed A1CF as a potential paralogue of RBM47, thus pointing to a possible role of RBM47 for C to U RNA editing. Rbm47 is co-expressed with Apobec1 and A1cf in the epithelial cells of the intestine. RBM47 colocalises with APOBEC1 and A1CF in the nucleus, where RNA editing takes place.²² RBM47 binds Apob mRNA and can interact with APOBEC1 and A1CF via the non-RRM region.¹⁷ In mouse 3T3 fibroblasts (which are inherently void of C to U RNA editing activity) or in vitro, RBM47 was able to substitute A1CF to mediate APOBEC1-editing of Apob RNA.¹⁷ In Rbm47^gt/gt mice that are deficient for Rbm47 due to a gene-trap (gt) mutation, C to U editing of Apob RNA is impaired.¹⁷ Taken together, these findings show that RBM47 is a new and essential component of the C to U editing machinery.

A new look for the core components of the C to U editosome

In addition to APOBEC1 and A1CF, the other known components of the editosome are the RBPs KHSRP,¹³ HNRNPAB (also known as ABBP1),²³ SYNCRIP (GRY-RBP; HNRNPQ),^15,24 HNRNPC1 ²⁵ and CELF2 (CUGBP2),¹⁶ and the co-chaperones DNAJB11 (ABBP2) ²⁶ and BAG4.²⁷ They were identified by their ability to interact with APOBEC1 or Apob RNA and modulate C to U editing.^12-16,23-27 However, in contrast to APOBEC1 and A1CF, these factors inhibit C to U RNA editing.²⁸ The discovery of RBM47 has added a new component to the C to U editosome. More importantly, it is a component that is required for editing and is sufficient on its own to mediate APOBEC1-editing activity in vitro,¹⁷ a property that is reputed to be unique to A1CF.¹⁰ RBM47 can substitute A1CF for C to U RNA editing.¹⁷ That either A1CF or RBM47 may fulfil the role of a requisite cofactor suggests that the constitution of the core machinery for C to U RNA editing is less stringent than previously anticipated. Furthermore, the potential existence of 2 different core editosomes raises many questions: Does the binding of RBM47 to Apob RNA, like for A1CF,¹² involve the essential 11-nucleotide mooring motif downstream of the editing site? Does RBM47 interact with other factors of the editosome? Are these 2 core editosomes performing redundant, complementary or specific functions? Are they acting on same or different RNA targets? Is there any other cofactor still to be discovered that could substitute for A1CF or RBM47 and constitute another core editosome?

In conjunction with APOBEC1, either RBM47 or A1CF is sufficient to elicit editing of Apob mRNA but together they do not produce any additive or synergistic effect, signifying that any interaction between these cofactors is not critical for editing.¹⁷ These findings suggest the possibility of functional redundancy between the 2 cofactors. However, while RBM47 is essential for RNA editing in mouse, the role of A1CF in vivo is unclear.²⁹ A1cf^−/− homozygous mutant mice die early during gestation, which preclude a proper analysis of its function. A1cf^+/− heterozygous mice are viable and, contrary to the requirement of A1CF for RNA editing in vitro, genetic haplo-insufficiency leads to an increase, rather than a reduction, in the level of editing of Apob RNA in the liver and the kidney (in which C to U RNA editing also takes place in rodents) relative to wild type level.²⁹ In the small intestine of the A1cf^+/− heterozygous mice, almost all Apob RNA is edited as in the wild type mice, thus making it impossible to detect any enhancement of the C to U editing activity. Apob RNA editing is unchanged or slightly reduced in the small intestine or the liver of the heterozygous Rbm47^+/gt mice and is abolished in Rbm47^gt/gt homozygous mutant mice (¹⁷ and unpublished data). Therefore, RBM47 is a necessary cofactor for C to U RNA editing in vivo while the role of A1CF remains uncertain. In this context, it is imperative to examine the level of Apob mRNA editing in conditional A1cf mutant mice that are not inflicted with early embryonic lethality. Similarly, it is necessary to test the role of additional putative candidate for C to U RNA editing as well as estimate the level of Apob RNA editing in mice that are mutated for other known components of the editosome in order to fully appreciate the contribution of each factor to this process in vivo.

Other putative factors for C to U RNA editing

The identification of RBM47 as a RNA editing factor stems from the discovery of sequence homology with a known C to U editing factor, A1CF.¹⁷ DND1, another putative paralogue of A1CF, can interact with APOBEC1.³⁰ Dnd1^ter mutant mice develop testicular germ cell tumors (TGCT),³¹ the incidence of which is modulated by the dosage of Apobec1-null allele.³² Apobec1 has been shown to affect TGCT susceptibility in a Mendelian manner, as well as via a parent-of-origin trans-generational epigenetic manner whereby a maternally inherited Apobec1 mutation can affect the tumorigenic susceptibility of 2 generations of offspring regardless of their Apobec1 genotype.³² Non-Mendelian bias in the transmission of the Apobec1-null and the Dnd1^ter alleles were also observed which is reminiscent of the preponderance of heterozygous A1cf^+/− individuals obtained from the crossings of A1cf mutant mice.²⁹ The molecular mechanism underpinning this pattern of inheritance and the involvement of RNA editing in this process is presently unknown but these observations highlight a functional connection between DND1, A1CF and APOBEC1. DND1 therefore appears as an obvious candidate for which it would be interesting to examine the role in C to U RNA editing.

APOBEC1 is part of the APOBEC family that can be divided in 5 sub-groups (AICDA, APOBEC1, APOBEC2, APOBEC3 and APOBEC4) containing 11 members in human and 5 members in mouse that are characterized by the presence of at least one zinc-dependent catalytic domain necessary for cytidine deamination.^33,34 Among them, human APOBEC3G can interact with HNRNPC1, HNRNPAB, SYNCRIP and HNRNPR.^35,36 In the mouse, APOBEC3 can interact with DND1 to counter its inhibitory activity on miRNA function.^30,37 HNRNPC1, HNRNPAB, SYNCRIP, HNRNPR and DND1 are involved in C to U RNA editing and/or related to A1CF.^10,21 While it is tempting to speculate that APOBEC3 proteins and other paralogues of APOBEC1 may similarly engage with C to U RNA editing, the studies so far have failed to observe such activity for these deaminases.^33,34 However, this may be because this was investigated with Apob, until recently the only RNA known to be edited, whereas the deaminases may act on other targets yet to be identified.

C to U editing beyond Apob mRNA

Apob is a natural RNA target of APOBEC1 editing, while 2 other RNA species, Nat1 and Nf1, can be edited under non-physiological condition.^38,39 32 new editing sites in 29 new RNA targets have been identified recently by pinpointing single-nucleotide mismatches in the RNA-sequencing reads of the wild type mice (compared to the reference genome: T instead of C or Adenine [A] instead of Guanine [G] for the negative strand transcripts) that were absent in the reads of the Apobec1^−/− mutant mice.¹⁸ An APOBEC1-editing sequence pattern was discerned in these new APOBEC1-edited RNA, which was subsequently detected in about 400 locations in the mouse genome. Analysis of some of the transcripts from these locations identified 9 additional C to U edited sites.¹⁸ In another study also of the Apobec1^−/− mutant mice, more than 40 extra edited sites were identified.¹⁹ We analyzed the editing status of the sites localized in Sult1d1, Serinc1, Casp6 and 2010106E10Rik mRNA in the Rbm47^gt/gt mutant mice. This revealed that the editing of all of them was impaired, suggesting that RBM47 may have a wide coverage of the transcriptome for APOBEC1-mediated editing.¹⁷ Blanc et al. tested some of these new sites and showed that contrary to Apob RNA, recombinant APOBEC1 and A1CF were not sufficient to mediate their editing.¹⁹ It will therefore be interesting to determine if RBM47 can bring about the editing of these new sites.

While these discoveries suggest that the function of C to U RNA editing is not only limited to the control of APOB48 protein expression and the formation of chylomicrons,⁴⁰ the functional consequence of the editing of these new sites is not known. Of particular interest though is the finding that almost all of these sites are located in the 3’ untranslated region (UTR) of mRNA. An attractive hypothesis is that C to U editing modifies the miRNA binding motif that influences the stability of the mRNA.¹⁸ However, the study by Blanc et al. did not find any association of the editing of the miRNA binding sites and the stability of the mRNA, but showed an effect on mRNA translation instead.¹⁹ As C to U RNA editing takes place concomitantly with splicing and poly-adenylation of the mRNA,²² another intriguing possibility is that editing could generate a AAUAAA motif,¹⁸ a recognition signal known for initiating cleavage and polyadenylation of pre-mRNA.⁴¹

A role beyond C to U RNA editing

APOBEC1 can bind AU-rich elements to stabilize RNA (e.g. IL8 mRNA ^42,43), a function that seems to be independent of its editing activity.^42-45 The association of APOBEC1 with IL8 mRNA requires SYNCRIP,⁴³ another factor that interacts with APOBEC1 to modulate its editing activity.^15,24 Interestingly, SYNCRIP codes different protein isoforms and the isoform that modulates C to U RNA editing may be different from the one that stabilises IL8 mRNA.⁴³ Apobec1^−/− mice on certain genetic background or maintained on specific diets show variable susceptibility to intestinal tumors and gallstone formation, which may be related to the ability of APOBEC1 to control the stability of Cox2 and Cyp7a1 mRNA.^44,45 A1CF and CELF2, another modulator of C to U editing,¹⁶ also bind the AU-rich region of Cox2 mRNA, and this leads to the stability of the mRNA in the case of CELF2.^10,46,47 In a similar context, the binding of RBM47 to intron and 3’ UTR may modulate splicing and/or stabilize different RNA species, including IL8 mRNA, to prevent tumor progression.⁴⁸ In addition, while RBM47, A1CF and APOBEC1 can interact with each other,¹⁷ their individual loss in mice has different phenotypic effect ^6-8,17,29 suggesting that they have different functional attributes.

Conclusion

The identification of novel RNA targets of APOBEC1-mediated editing, and the discovery of RBM47 that can act as a core component of the editosome in place of A1CF have garnered renewed interest in C to U RNA editing. The proposition of the common functionality of the potential paralogues of A1CF and APOBEC1 for this process is especially appealing. Future works will test the possibility that the association with different core components may underpin the target specificity of the editing machinery. At this stage, another important task is to characterize the biological role of the editing of the novel RNA targets. It would also be interesting to find out if the event of C to U RNA editing, like that of Adenine (A) to Inosine (I) editing,⁴⁹ can impact on a diverse range of biological processes, e.g., cell-cell signaling, growth and homeostasis, and tumorigenesis.⁵⁰

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Karin Tourle for critical reading of the manuscript.

Funding

Our work is supported by the Australian Research Council (Grant DP130100779) and Mr James Fairfax. NF is the Sir Norman Gregg Research Fellow of CMRI. PPLT is a NHMRC Senior Principal Research Fellow (Grant 1003100).