Dcp2: an mRNA decapping enzyme that adopts many different shapes and forms.

Eukaryotic mRNAs contain a 5' cap structure that protects the transcript against rapid exonucleolytic degradation. The regulation of cellular mRNA levels therefore depends on a precise control of the mRNA decapping pathways. The major mRNA decapping enzyme in eukaryotic cells is Dcp2. It is regulated by interactions with several activators, including Dcp1, Edc1, and Edc3, as well as by an autoinhibition mechanism. The structural and mechanistical characterization of Dcp2 complexes has long been impeded by the high flexibility and dynamic nature of the enzyme. Here we review recent insights into the catalytically active conformation of the mRNA decapping complex, the mode of action of decapping activators and the large interactions network that Dcp2 is embedded in.

Current Opinion in Structural Biology 2019, 59:115–123

This review comes from a themed issue on Protein nucleic acid interactions

Edited by Frédéric H-T Allain and Martin Jinek

For a complete overview see the and the

Available online 29th August 2019

0959-440X/© 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Eukaryotic mRNA is co-transcriptionally modified with a N7-methylgaunosine (m7G) cap at the 5’ end (Figure 1a). This cap is essential for translation initiation and interacts with the initiation factor eIF4F that recruits the small ribosomal subunit. eIF4F also interacts with the polyA binding protein (PABP) that associates with the 3’ polyA tail of the mRNA and thereby links the 5’ and 3’ ends of the mRNA [1]. Recent results suggest that the formation of this closed loop structure can be modulated to control mRNA stability [2,3]. The 5’ cap also protects mRNAs from degradation by 5’-3’ exonucleases and thereby enhances the lifetime of the transcript [4].

Figure 1

Overview of Dcp2 mediated mRNA decapping. (a) Structure of m7G capped mRNA. The Dcp2 mediated decapping reaction produces m7GDP and 5’-monophosphate mRNA (b) 5’-3’ mRNA decay pathway: Shortening of the polyA tail by the CCr4–Not complex and Pan2/3 leads to dissociation of the PABP and recruitment of the LSm1-7:Pat1 complex. The interaction between Pat1 and the IDR of Dcp2 facilitates binding of the Dcp1:Dcp2 complex. After decapping of the mRNA by Dcp2, the exonuclease Xrn1 is recruited and rapidly degrades the mRNA in the 5’-3’ direction. (c) Domain organization of the S. pombe Dcp1:Dcp2 complex.

The removal of the 5’ cap structure from the mRNA is an important control step during bulk mRNA degradation [5] as well as in nonsense-mediated decay [6], AU-rich mRNA decay [7] and miRNA-mediated mRNA turnover [8]. Bulk decay in the 5’-3’ direction starts with the shortening of the 3’ polyA tail by the CCR4–Not and Pan2/3 complexes (Figure 1b) [5]. This leads to dissociation of the PABP, after which the LSm1-7:Pat1 complex can interact with the shortened 3’ polyA tail. Pat1 then recruits the Dcp1:Dcp2 decapping complex (Figure 1c). Dcp2 [9, 10, 11], the catalytic subunit of the decapping complex, is then able to hydrolyzes the m7G cap. This decapping reaction generates a 5’-monophosphate mRNA and m7GDP (Figure 1a) [10,11]. After decapping, the mRNA body is rapidly degraded in the 5’-3’ direction by the exonuclease Xrn1 [5]. Decapping was long assumed to be an irreversible step during mRNA decay; however, there is initial evidence that recapping can take place in higher eukaryotes [12].

The Dcp2 decapping enzyme

The Dcp2 enzyme is the major decapping enzyme in 5’-3’ mRNA decay (Figure 1c) [4]. Dcp2 consists of an N-terminal regulatory domain (RD), followed by the catalytic domain (CD) and an intrinsically disordered C-terminal tail (IDR) that varies between around 100 residues in plants and humans to over 500 residues in yeast. The CD belongs to the ubiquitous NUDIX hydrolase family that generally catalyzes the hydrolysis of diphosphates linked to nucleosides [10,11,13]. Catalysis is performed by the NUDIX motif, a loop-helix-loop structure that coordinates catalytic Mg2+ ions via three conserved glutamates. NUDIX hydrolases often possess a broad substrate specificity and the identification of their physiological substrate is sometimes challenging [14]. It is thus not surprising that several other NUDIX hydrolases have been identified that are able to remove the m7G cap from RNAs in vitro (e.g. Nudt2/3,12,15-17/19) [15], but of those, Dcp2 is the only enzyme that shows a clear specificity for m7G capped RNA and that interacts with other mRNA decay factors.

Here we review recent insights into the structural basis behind the regulation and specific cap recognition of the Dcp2 enzyme. Most of the structural and functional data have been obtained for two yeast model system (Schizosaccharomyces pombe and Kluyveromyces lactis) and we will focus on those results. Many of the principles are likely conserved in higher eukaryotes, but preliminary data show that a number of the interactions sites in the mRNA decay network have been reshuffled during evolution [16,17].

The activity of Dcp2 is enhanced by multiple mRNA decay factors

The low basal mRNA decapping activity of the isolated CD of Dcp2 is greatly enhanced by several consecutive mechanisms. First, the N-terminal RD of Dcp2 enhances the catalytic activity of the CD by about two orders of magnitude [18,19,20]. Second, the Dcp2 RD tightly interacts with the main decapping activator Dcp1 (Figure 1c), which results in additional enhancements in decapping activity in vitro and in vivo [9,21]. Dcp1 belongs to the EVH1 domain family of proteins and is able to use a dynamic, hydrophobic β-sheet surface to recruit additional mRNA decay factors that contain proline-rich motifs [22,23]. These additional decapping factors include Edc1 and Edc2 that have been shown to further increase the Dcp1:Dcp2 decapping activity in vivo and in vitro [9].

The Dcp2 enzyme displays multiple domain orientations

How Dcp2 specifically recognizes the 5’ m7G cap structure and the mechanisms behind decapping enhancement have long been unresolved questions due to the flexible nature of the Dcp2 enzyme. Indeed, the RD and CD in Dcp2 are connected by a short (∼4 aa) flexible linker and both domains undergo rapid transitions between an open state without interdomain contacts and a closed state, where both domains interact [20,24,25]. Early work already indicated that a particular interdomain orientation is a prerequisite for efficient catalysis and that the m7G cap structure is recognized by a “split active site” that comprises residues from the RD as well as the CD [26,27].

The active state of Dcp2

Over the years, several crystal structures of Dcp2 in the apo state as well as in complex with Dcp1, decapping enhancers Edc1 and Edc3, substrate analogs and the mRNA decapping product m7GDP have been solved (Figure 2a) [18,20,24,28,29,30,31]. In line with the observed dynamics, the orientation between the Dcp2 domains is vastly different in these structures (Figure 2a). A principle component analysis reveals that these static structures can be grouped into six different clusters with different interdomain orientations between the RD and CD of Dcp2. No clear correlation between the complex composition and the domain orientation is evident, preventing conclusions that link the known structures to the catalytic cycle of Dcp2 [32,33]. In addition, in most of these structures the catalytic NUDIX helix points away from the RD, which is not compatible with the recognition of the mRNA cap by the RD in a split active site [27] and the known mechanisms of NUXIX hydrolases.

Figure 2

Structures of the Dcp2 enzyme. (a) Crystal structures of Dcp2 in isolation and in different complexes. The orientation of the Dcp2 RD (dark blue) is identical in all structures and the structures are grouped according to the orientation between RD and CD (yellow) of Dcp2. The catalytic NUDIX helix is shown in red, bound ligands in green, Edc1 and Edc3 in pink and Dcp1 in light blue. References and PDB codes are shown below the structures. (b) The split active site in the catalytically competent orientation 6a. The cap structure is recognized by W43 and D47 in the RD and by R190 and K191 in the CD. The catalytically important Mg2+ ions (green spheres) are bound by the NUDIX helix and come close to the triphosphate linkage. (c) The active conformation in the Dcp1:Dcp2:Edc1:m7GDP complex (orientation 6a). m7GDP and the YAG activation motif of Edc1 are sandwiched between the CD and RD of Dcp2. The proline rich region in Edc1 interacts with Dcp1. (d) The RNA body binds to a positively charged surface in the active conformation (conformation 6c). Dcp2 is colored according to the electrostatic surface potential (blue positive, red negative, other proteins are colored as in (a)). The RNA binding path is indicated in green. The structure of the two-headed cap analog used for crystallization is shown at the bottom.

The only the Dcp2 domain orientation that is in agreement with a catalytically competent form of the enzyme is adopted in the three structures with orientation 6 (Figure 2). These key structures include (i) the S. pombe Dcp1:Dcp2:Edc1:m7GDP complex [20], (ii) the K. lactis Dcp1:Dcp2:Edc3:m7GDP complex [30] and iii) the S. pombe Dcp1:Dcp2:Edc1:Edc3 proteins in complex with a nonhydrolyzable cap analog that contains a tetraphosphate linker [31]. In those structures, the m7G cap is sandwiched between W43 (residue numbering refers to the proteins from S. pombe) in the RD and R190 and K191 in a conserved loop of the CD, in agreement with the predicted split active site (Figure 2b). In addition, the Watson–Crick edge of the m7G cap is recognized by two hydrogen bonds to D47 in the RD. Finally, the phosphate groups that link the m7G cap with the RNA body are bound to the NUDIX motif via 3 Mg2+ ions (Figure 2b). The cap binding site in orientation 6 is also in excellent agreement with NMR titration experiments [27] and the known mechanism of NUDIX hydrolases [13].

Two of the structures in orientation 6 also contain the decapping activator Edc1. Edc1 is a 200 aa long intrinsically disordered protein, but a fragment of ∼25 aa that contains a YAGxxF activation motif followed by a proline rich region is sufficient for decapping activation [22,23,28]. The proline rich region in Edc1 binds to the β-sheet surface on Dcp1, whereas the YAG sequence of the activation motif is sandwiched between the Dcp2 RD and CD and forms several stacking and hydrogen bonding interactions with both domains (Figure 2c). The catalytically active conformation of Dcp2 is thus stabilized by both the YAG sequence in Edc1 and the m7G cap structure, as both factors specifically bridge between the two Dcp2 domains.

Interestingly, the structure of the active conformation was also determined in the presence of a two headed cap analog (Figure 2a, conformation 6c) [31]. In that structure the first base of the mRNA body stacks onto a conserved aromatic residue (Y220 in S. pombe; F223 in K. lactis) of the CD below the cap binding site (Figure 2d). This residue is important for substrate recognition as mutations abolish the preference of Dcp2 for mRNAs containing a purine residue at the first position [31]. The binding surface for the remaining of the mRNA body is formed by a positively charged surface patch that starts below the cap binding site and extends toward the C-terminal region of the CD (Figure 2d). In the active conformation of the enzyme this binding groove is fully exposed.

In summary, the Dcp2 structures that adopt orientation 6 rationalize a plethora of previously published biochemical data and thus represent the catalytically active conformation of the Dcp1:Dcp2 complex. Nevertheless, based on these static crystal structures it is not possible to draw conclusion on the conformation of Dcp2 and its complexes with Dcp1, decapping activators or mRNA in solution. This is also evident from the fact that 3 crystal structures with different domain orientations (orientations 2, 3, 4a) have been solved for the Dcp1:Dcp2 complex.

Combining solution state data with static structures

Solution-based methods, including small angle scattering and solution NMR spectroscopy are invaluable to correlate static crystal structures with conformations that are adopted in solution. These methods provide information regarding the overall shape, respectively the dynamics and structure of the complex in solution [34]. Recently, NMR spectroscopic methods were exploited to determine the conformation of Dcp2 in solution and to address the influence of decapping activators on the conformation (Figure 3) [20].

Figure 3

Conformations that Dcp2 adopts in solution. Free Dcp2 (top left) equally populates a dynamic, open state (grey) and the closed state (conformation 4, orange). Binding of Dcp1 stabilizes the closed state (top middle), whereas binding of Edc1 has no influence on the conformations of Dcp2 (top right). Binding of capped RNA to the Dcp1:Dcp2:Edc1 complex locks Dcp2 in the active state (bottom right, conformation 6, red). Addition of capped mRNA to Dcp1:Dcp2 in the absence of Edc1 is not sufficient to lock Dcp2 in the active state but competes with the closed state and leads to a mixture of open and active state (bottom middle).

Interestingly, the isolated Dcp2 enzyme interconverts rapidly between two equally populated states (lifetime ∼1 ms) [20,25] (Figure 3). In accordance with paramagnetic relaxation enhancement NMR experiments [35] these conformational states were linked to the closed conformation observed in orientation 4a–c (Figures 2, 3) and to an open conformation without any interdomain contacts, as displayed in orientation 3.

Binding of Dcp1 to the RD of Dcp2 domain shifts the Dcp2 open-closing equilibrium significantly toward the closed state with little influence on the exchange rate [20]. Interestingly, this closed state is catalytically incompetent as the split active site is not formed and the RNA binding site on the CD is blocked by the RD. In line with this the Dcp1:Dcp2 complex shows a reduced RNA affinity compared to Dcp2 in isolation [20].

In solution, the stable catalytically active conformation (orientation 6) is only adopted in the presence of Edc1 and substrate. In this Dcp1:Dcp2:Edc1:mRNA complex the YAG activation motif of Edc1 and the m7G cap enforce the formation of the split active site (see above) in a synergistic manner. Neither substrate alone, nor Edc1 alone is able to force Dcp2 into the stable active conformation. Indeed, the Dcp1:Dcp2:Edc1 complex samples the same orientations as the Dcp1:Dcp2 complex and the Dcp1:Dcp2:mRNA complex in the absence of Edc1 samples mainly open conformations, as mRNA binding competes with the closed state and the active state is formed only transiently.

In summary, only three out of six crystallized Dcp2 interdomain orientations (orientation 3, 4, and 6) have been detected in solution. This highlights the need for integrative structural biology approaches and especially solution methods, when dealing with dynamic multi-domain complexes [36]. In that light, it is also important to note that mutational approaches that are designed to validate the Dcp2 structures have been misleading. As an example, the mutation of W43 and D47 in the RD have strong effects on catalytic activity [18]. These residues are at the interface between the RD and CD in the closed (inactive) form of the enzyme (orientation 4) and initially these findings were interpreted as an indication for the importance of this closed state in catalysis [24]. However, these residues were later found to be directly involved in the recognition of the mRNA cap in the active state of the enzyme.

Finally, the crystal structures that display Dcp2 domain orientations that are not observed in solution (orientation 1, 2 and 5) are likely artificially induced by the crystal lattice; future structural work will reveal if specific mRNA decapping factors are able to stabilize these catalytically inactive orientations in solution.

Enhancement of the activity

Importantly, the structure of the active form of Dcp2 in combination with the solution data rationalize how the low catalytic selectivity and activity of the isolated CD domain is increased in a stepwise manner by the RD, Dcp1, and Edc1 (see above). First, the RD increases the activity and selectivity of the CD by completing the cap binding site. Interestingly, the isolated CD hydrolyzes 5’-triphosphate RNAs faster than 5’ capped RNAs; after inclusion of the RD domain the decapping activity is significantly stimulated, whereas the 5’ triphosphate activity is not influenced by the RD [20]. This argues for a scenario where the RD has been linked to the CD during evolution in order to enhance selectivity for the mRNA cap structure that appeared in eukaryotes. The second step in the activation of Dcp2 is through the recruitment of Dcp1. Mechanistically, this activation is achieved through the stabilization of the fold of the RD domain [20]. Finally, the activation of Dcp1:Dcp2 through Edc1 is mediated by the YAG motif that stabilizes the active conformation in the presence of substrate.

Interestingly, two of the Dcp2 crystal structures (conformations 6b and 6c) were solved in the presence of the LSm domain of the decapping activator Edc3 that is able to increase the activity of Dcp2 by around 30% [17,37]. The LSm domain binds to helical leucine-rich motifs (HLMs) that are located in the IDR C-terminal to the CD. These HLMs fold into an amphipathic alpha-helix in the bound state [17], which leads to an extension of the C-terminal alpha-helix of the CD [30,31] and an enlargement of the RNA-binding site in Dcp2. The recruitment of Edc3 to the Dcp2 complex has no direct influence on the Dcp2 domain orientation in solution (P. Wurm and R. Sprangers, unpublished data) and thus increases the decapping activity through a mechanism that is fundamentally different from one used by the activator Edc1.

The interaction network that involves Dcp2

As decapping primes an mRNA transcript for degradation the activity of the Dcp2 enzyme has to be tightly controlled. To that end, the complex is embedded in a large interaction network encompassing many mRNA decay factors (Figures 1, 4) [38, 39, 40, 41, 42]. Many of these interactions involve the IDR of Dcp2, which directly interacts with the decapping activators Pat1, Upf1 and Edc3 via so called short linear motifs (SLiMs) [16,43]. These are short intrinsically disordered regions that bind to folded protein domains. This kind of interaction is susceptible to rapid rearrangement during evolution. As a result some of the SLiMs in Dcp2 have been transferred to a disordered C-terminus that is present in Dcp1 from metazoa [16,17]. Interestingly, the IDR of Dcp2 harbors multiple binding sites for Upf1 (at least 2) [43], Edc3 (at least 7) [17,44] and Pat1 (at least 8) [45]. Recently, two structures of the C-terminal alpha-helical domain of Pat1 in complex with two of these Dcp2 SLiMs have been solved [45] (Figure 4). This revealed that Pat1 and Edc3 bind to similar helical leucine-rich motifs and thus potentially compete for binding to Dcp2 [45]. It will be interesting to see how the Dcp2 IDR and the helicase Upf1, an important factor in nonsense-mediated mRNA decay, interact in detail.

Figure 4

The Dcp1:Dcp2 complex is part of a dense interaction network of mRNA decay factors. Known interaction partners of Dcp1:Dcp2 are shown and the interactions between them are indicated. Interactions are colored according to the interaction partners (black: interactions between folded domains, red: interactions involving a SLiM, grey: interaction sites are unknown). Many of the interactions are mediated by SLiMs and known structures of SLiMs bound to folded domains are shown (SLiMs are colored red, folded domains orange, PDB codes are shown below the structures). Note that the Pdc1 protein is not present in all yeast species.

In addition to its function as a binding hub for intermolecular interactions, the Dcp2 IDR contains two autoinhibitory motifs that interact with the Dcp2 core domains [46]. Experimental evidence suggest that this inhibition is achieved via stabilization of the closed, catalytically inactive conformation of Dcp2 [46]. This Dcp2 autoinhibition is abrogated by Edc3 and potentially by Pat1, as these proteins bind close to the Dcp2 autoinhibitory elements.

In summary, the intrinsically disordered C-terminus of Dcp2 allows for the redundant recruitment of multiple proteins that are involved in mRNA decay. In line with that, a recent high-throughput study found that deletion of the C-terminal tail of Dcp2 leads to specific upregulation as well as downregulation of several hundred mRNA transcripts [47].

Additional decapping factors can be recruited to the Dcp2 enzyme through interactions with Dcp1 (Figure 4). This includes Edc1, as shown above, but also the mRNA decay factors Dhh1, Pat1 and XrnI and the autoinhibitory region in Dcp2, that all contain proline rich regions [22,48]. Most of these interactions are of medium to low affinity; however, avidity effects due to the large number multivalent of interactions are likely to enhance the interaction strength.

Processing bodies

Many of the intermolecular interactions that the Dcp1:Dcp2 complex is involved in contribute to the formation of processing bodies (P-bodies) [11,49,50]. These are cytoplasmic ribonucleoprotein foci that contain translationally repressed mRNAs and mRNA decay factors (including Dcp2, Dcp1, Edc3, Pat, LSm1-7, and Dhh1) [51]. P-body formation is the result of liquid-liquid phase separations [52] which depend on multivalent protein–protein and protein–RNA interactions and on low-complexity protein-sequences [44,53]. The cellular function of P-bodies is not entirely clear, initially they were regarded as the sites of cellular mRNA degradation, but there is accumulating evidence that they also serve as storage sites for translationally repressed mRNAs, which can later be degraded or reenter translation [51]. In line with this, the catalytic activity of Dcp2 enzyme is decreased upon phase separation [54].

In summary, we have here reviewed the recent progress in our understanding of the structural basis of decapping by Dcp2 and its regulation by mRNA decay factors. Despite this progress many questions remain open, including the mechanism by which the Dcp2 enzyme is inhibited by its transition into processing bodies, by interactions with the autoinhibitory elements and by m6A methylation of the first mRNA base [55]. In addition, it remains unclear how interactions between Dcp2 and Upf1 results in the activation of the NMD pathway as well as how the Dcp2 enzyme is structured and regulated in higher eukaryotes. Insights into these questions might also shed light on how mRNA selectivity in the decapping process is achieved. We are looking forward to future results that address these exciting questions.

Conflict of interest statement

Nothing declared.

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

• of special interest

•• of outstanding interest