RNA modifications and structures cooperate to guide RNA-protein interactions

Abstract

An emerging body of evidence indicates that post-transcriptional gene regulation relies not only on the sequence of mRNAs but also on their folding into intricate secondary structures and on the chemical modifications of the RNA bases. These features, which are highly dynamic and interdependent, exert direct control over the transcriptome and thereby influence many aspects of cell function. Here, we consider how the coupling of RNA modifications and structures shapes RNA-protein interactions at different steps of the gene expression process.

FIGURE 1. RNA Modifications and Their Biological Functions

a | Chemical structures of unmodified RNA bases (top) and the modified structures (below). b | Table of writers, readers, and erasers of RNA modifications. c | Biological effects of RNA modifications. In response to stress, N¹-methyladenosine (m¹A) and N⁶-methyladenosine (m⁶A) levels in 5′UTRs are dynamically altered. m¹A promotes cap-dependent translation through an unknown reader, while 5′UTR m⁶A facilitates cap-independent translation through interactions with eIF3. 5′UTR m⁶A is protected from demethylation by YTHDF2. In coding sequences, 5-hydroxymethylcytosine (hm⁵C) increases the translation efficiency of modified transcripts via an unknown mechanism. hm⁵C levels are highest in the brain, and ablation of TET genes affects brain development in Drosophila melanogaster. m⁶A in coding regions affects RNA processing by modulating the binding of YTHDC1, hnRNPC, hnRNPA2B1, and SRSF2. Removal of m⁶A by FTO antagonizes SRSF2 binding and is necessary for adipogenesis. 3′UTR m⁶A also affects RNA processing via the aforementioned readers, as well as cap-dependent translation via YTHDF1 and RNA degradation via YTHDF2. Regulation of RNA stability by m⁶A is crucial for stem cell differentiation and circadian clock control. 5-methylcytosine (m⁵C) is linked to translational control of senescence-related genes. Pseudouridine (ψ) in 3′ UTRs has been shown to increase the stability of target transcripts during heat shock. Gray arrows are used for modifications with no known readers. Red arrows connect readers to biological processes they are directly associated with.

FIGURE 1. RNA Modifications and Their Biological Functions

a | Chemical structures of unmodified RNA bases (top) and the modified structures (below). b | Table of writers, readers, and erasers of RNA modifications. c | Biological effects of RNA modifications. In response to stress, N¹-methyladenosine (m¹A) and N⁶-methyladenosine (m⁶A) levels in 5′UTRs are dynamically altered. m¹A promotes cap-dependent translation through an unknown reader, while 5′UTR m⁶A facilitates cap-independent translation through interactions with eIF3. 5′UTR m⁶A is protected from demethylation by YTHDF2. In coding sequences, 5-hydroxymethylcytosine (hm⁵C) increases the translation efficiency of modified transcripts via an unknown mechanism. hm⁵C levels are highest in the brain, and ablation of TET genes affects brain development in Drosophila melanogaster. m⁶A in coding regions affects RNA processing by modulating the binding of YTHDC1, hnRNPC, hnRNPA2B1, and SRSF2. Removal of m⁶A by FTO antagonizes SRSF2 binding and is necessary for adipogenesis. 3′UTR m⁶A also affects RNA processing via the aforementioned readers, as well as cap-dependent translation via YTHDF1 and RNA degradation via YTHDF2. Regulation of RNA stability by m⁶A is crucial for stem cell differentiation and circadian clock control. 5-methylcytosine (m⁵C) is linked to translational control of senescence-related genes. Pseudouridine (ψ) in 3′ UTRs has been shown to increase the stability of target transcripts during heat shock. Gray arrows are used for modifications with no known readers. Red arrows connect readers to biological processes they are directly associated with.

FIGURE 1. RNA Modifications and Their Biological Functions

a | Chemical structures of unmodified RNA bases (top) and the modified structures (below). b | Table of writers, readers, and erasers of RNA modifications. c | Biological effects of RNA modifications. In response to stress, N¹-methyladenosine (m¹A) and N⁶-methyladenosine (m⁶A) levels in 5′UTRs are dynamically altered. m¹A promotes cap-dependent translation through an unknown reader, while 5′UTR m⁶A facilitates cap-independent translation through interactions with eIF3. 5′UTR m⁶A is protected from demethylation by YTHDF2. In coding sequences, 5-hydroxymethylcytosine (hm⁵C) increases the translation efficiency of modified transcripts via an unknown mechanism. hm⁵C levels are highest in the brain, and ablation of TET genes affects brain development in Drosophila melanogaster. m⁶A in coding regions affects RNA processing by modulating the binding of YTHDC1, hnRNPC, hnRNPA2B1, and SRSF2. Removal of m⁶A by FTO antagonizes SRSF2 binding and is necessary for adipogenesis. 3′UTR m⁶A also affects RNA processing via the aforementioned readers, as well as cap-dependent translation via YTHDF1 and RNA degradation via YTHDF2. Regulation of RNA stability by m⁶A is crucial for stem cell differentiation and circadian clock control. 5-methylcytosine (m⁵C) is linked to translational control of senescence-related genes. Pseudouridine (ψ) in 3′ UTRs has been shown to increase the stability of target transcripts during heat shock. Gray arrows are used for modifications with no known readers. Red arrows connect readers to biological processes they are directly associated with.

FIGURE 2. Stress Induces N⁶-methyladenosine Driven Cap-Independent Translation

Under normal conditions (left), N⁶-methyladenosine (m⁶A) within the 5′UTRs of stress-responsive transcripts is removed by the eraser fat mass and obesity-associated (FTO). Upon export to the cytoplasm, the unmodified transcripts undergo efficient cap-dependent translation, whereas the demethylated transcripts encoding stress-responsive genes are minimally translated. Upon heat shock (right), the m⁶A reader YTH domain family 2 (YTHDF2) translocates from the cytoplasm to the nucleus. Nuclear YTHDF2 then binds to 5′UTR m⁶A in the stress-responsive mRNAs and prevents them from FTO-mediated demethylation. These conditions inhibit cap-dependent translation, thus impairing protein production from unmodified transcripts. However, the m⁶A residues within the 5′ UTRs of stress responsive genes are directly bound by eukaryotic initiation factor 3 (eIF3), facilitating rapid cap-independent translation.

FIGURE 3. Impact of RNA Structures on Alternative Splicing

a | A majority of RNA binding protein, Fox 2 homolog (RBFOX2) binding sites are over 500 nucleotides away from the nearest exon. To allow splicing regulation from these distal binding sites, base-pairing interactions occur within the intron, which bring RBFOX2 into close proximity to its target exons. b | N⁶-methyladenosine (m⁶A) deposition destabilizes base-pairing interactions and disrupts local secondary structure. In doing so, m⁶A can expose previously buried RNA-binding motifs to RNA-binding proteins, such as the U-tracts bound by heterogeneous ribonucleoprotein C (HNRNPC). The figure is reproduced, with permission, from REF. . c | G-qaudruplexes (RGQs) form from the stacking of three or more G-quartets (top). In 5′ untranslated regions (UTRs), RGQs impede the initiation and scanning steps of translation. In 3′UTRs, RGQs can block access to microRNA (miRNA) binding sites (shown in red), thus preventing miRNA-mediated decay and translational repression. These structures are dynamic and can be unwound by RNA helicases.

FIGURE 3. Impact of RNA Structures on Alternative Splicing

a | A majority of RNA binding protein, Fox 2 homolog (RBFOX2) binding sites are over 500 nucleotides away from the nearest exon. To allow splicing regulation from these distal binding sites, base-pairing interactions occur within the intron, which bring RBFOX2 into close proximity to its target exons. b | N⁶-methyladenosine (m⁶A) deposition destabilizes base-pairing interactions and disrupts local secondary structure. In doing so, m⁶A can expose previously buried RNA-binding motifs to RNA-binding proteins, such as the U-tracts bound by heterogeneous ribonucleoprotein C (HNRNPC). The figure is reproduced, with permission, from REF. . c | G-qaudruplexes (RGQs) form from the stacking of three or more G-quartets (top). In 5′ untranslated regions (UTRs), RGQs impede the initiation and scanning steps of translation. In 3′UTRs, RGQs can block access to microRNA (miRNA) binding sites (shown in red), thus preventing miRNA-mediated decay and translational repression. These structures are dynamic and can be unwound by RNA helicases.

FIGURE 3. Impact of RNA Structures on Alternative Splicing

a | A majority of RNA binding protein, Fox 2 homolog (RBFOX2) binding sites are over 500 nucleotides away from the nearest exon. To allow splicing regulation from these distal binding sites, base-pairing interactions occur within the intron, which bring RBFOX2 into close proximity to its target exons. b | N⁶-methyladenosine (m⁶A) deposition destabilizes base-pairing interactions and disrupts local secondary structure. In doing so, m⁶A can expose previously buried RNA-binding motifs to RNA-binding proteins, such as the U-tracts bound by heterogeneous ribonucleoprotein C (HNRNPC). The figure is reproduced, with permission, from REF. . c | G-qaudruplexes (RGQs) form from the stacking of three or more G-quartets (top). In 5′ untranslated regions (UTRs), RGQs impede the initiation and scanning steps of translation. In 3′UTRs, RGQs can block access to microRNA (miRNA) binding sites (shown in red), thus preventing miRNA-mediated decay and translational repression. These structures are dynamic and can be unwound by RNA helicases.

FIGURE 4. Dynamics of Modification Fractions and Stoichiometry

Within the pool of transcripts from a given gene, only a fraction of each site is modified. For instance, under certain cellular conditions, a small percentage of the transcripts may contain a base modification (Cell State 1, left panel). By causing structural rearrangements and directly recruiting modification reader proteins, the modified pool of pre-mRNAs has a different cellular fate, which could result in increased turnover as shown (though alternate fates are possible). As cellular conditions change, modification stoichiometry may change as well, resulting in a higher (Cell State 2, right panel) or lower percentage of modified transcripts due to alterations in reader, writer, and eraser activities or expression.