Normal and Aberrantly Cap mRNA Decapping

Abstract

Messenger RNAs transcribed by RNA Polymerase II are modified at their 5′ end by the cotranscriptional addition of a 7-methylguanosine (m⁷G) cap. The cap is an important modulator of gene expression and the mechanism and components involve in it removal have been extensively studied. At least two decapping enzymes, Dcp2 and Nudt16, and an array of decapping regulatory proteins remove the m⁷G cap from an mRNA exposing the 5′ end to exonucleolytic decay. In contrast, relatively less is known about the decay of mRNAs that may be aberrantly capped. The recent demonstration that the Saccharomyces cerevisiae Rai1 protein selectively hydrolyzes aberrantly capped mRNAs provides new insights into the modulation of mRNA that lack a canonical m⁷G cap 5′-end. Whether an mRNA is uncapped or capped but missing the N7 methyl moiety, Rai1 hydrolyzes its 5′-end to generate an mRNA with a 5′ monophosphate. Interestingly, Rai1 heterodimerizes with the Rat1 5′−3′ exoribonuclease, which subsequently degrades the 5′-end monophosphorylated mRNA. Importantly, Rat1 stimulates the 5′-end hydrolysis activities of Rai1 to generate a 5′ end unprotected mRNA substrate for Rat1 and in turn, Rai1 stimulates the activity of Rat1. The Rai1-Rat1 heterodimer functions as a molecular motor to detect and degrade mRNAs with aberrant caps and defines a novel quality control mechanism that ensures mRNA 5′-end integrity. The increase in aberrantly capped mRNA population following nutritional stress in Saccharomyces cerevisiae demonstrates the presence of aberrantly capped mRNAs in cells and further reinforces the functional significance of the Rai1 in ensuring mRNA 5′end integrity.

II. Introduction

The stability and translational efficiency of eukaryotic mRNAs are significantly influenced by the 5′-end 7-methylguanosine (m⁷G) cap of eukaryotic mRNAs [1–3]. The cap is cotranscriptionally added and consists of a guanine nucleoside methylated at the N7 position attached to the terminal nucleoside of the RNA by an unusual 5′−5′ pyrophosphate linkage [4–6]. Capping is carried out by the combination of three enzymatic activities consisting of a triphosphatase, guanylyltransferase and methyltransferase [5, 7]. Three different proteins carry out the distinct activities in yeast while the triphosphatase and guanylyltransferase activities are carried out by a single bifunctional capping enzyme in mammals [8]. The presence of the methyl group on the cap is essential for recognition by the cap-binding complex, CBC and eIF4E [9–11]. The capping enzymes are recruited to the nascent transcript generated by RNA Polymerase II through its carboxyl terminal domain (CTD).

Historically, cap addition has been perceived as a default process that lacks regulation and proceeds to completion. In contrast, removal of the cap is where the regulation is believed to occur and catalyzed by the Dcp2 [12–14] and Nudt16 [15, 16] decapping enzymes to release m⁷GDP and 5’-monophosphate RNA. The exposed 5′ monophosphate RNA is subsequently subjected to degradation by the Xrn1 5′ to 3′ exoribonuclease to clear the mRNA body [17, 18]. Interestingly, Dcp2 and Nudt16 function on an N7 methyl cap substrate and minimally function on unmethylated cap [13, 16] raising an intriguing question regarding the fate of mRNAs aberrantly lacking a cap or the N7 methyl moiety. Recent reports have shown that at least in yeast cells, aberrantly capped mRNAs are generated upon exposure to nutrient stress and a heterodimeric complex consisting of Rai1-Rat1 degrades these mRNAs. Rai1 is an aberrant cap decapping enzyme that preferentially hydrolyzes mRNAs with noncanonical 5′ ends to generate a 5′ -end monophosphate mRNA that can be utilized as substrate for the nuclear Rat1 exoribonuclease. Importantly, the two subunits mutually stimulate one another’s activity whereby Rai1 enhances Rat1 activity and Rat1 stimulates Rai1 activity. Here, we discuss the contribution of Rai1 to aberrant cap decapping to initiate decay of mRNAs with aberrant 5′-ends in a 5′-end quality control mechanism.

III. mRNA decapping proteins in the Exonucleolytic pathway of mRNA decay

In eukaryotes, mRNA decay is generally triggered by shortening of the 3′ polyA tail to deadenylate the mRNA (please see Chapter 10 for a review of mRNA deadenylation). The deadenylated mRNA is subsequently shunted into one of two different directional decay pathways to undergo either 5′-to-3′ or 3′-to-5′ decay. In the 3′-to-5′ pathway, the RNA is degraded from the 3′end by the RNA exosome, which consists of a multisubunit complex with a 3′-to-5′ exoribonuclease component (please see chapter 4 for a review of the exosome). The resulting cap structure is hydrolyzed by a scavenger decapping enzyme, DcpS, to release m⁷GMP and nucleotide diphosphate [19, 20]. In the 5′-to-3′ decay pathway, the mRNA 5′ end cap structure is initially cleaved by the 5′ decapping enzymes, Dcp2 or Nut16, to release m⁷GDP and a 5′-end monophosphorylated RNA [16, 21–24] (Fig 1). Following removal of the cap, the resulting 5′ monophosphorylated RNA serves as a substrate for one of two major 5′−3′ exoribonucleases, the nuclear yeast Rat1 exoribonuclease (Xrn2 in mammals) or cytoplasmic Xrn1 protein [18, 25].

Fig 1. Eukaryotic exonucleolytic mRNA decay pathways.

Following shortening of the poly(A) tail, eukaryotic mRNAs can undergo either 5′-to-3′ or 3′-to-5′ decay. In the 5′ decay pathway, the mRNA 5’ end is decapped by Dcp2 or Nudt16 to release m⁷GDP and generates a 5′-end monophosphorylated mRNA that is subsequently degraded by exoribonucleases Xrn1 or Xrn2. In the 3′ decay pathway, the deadenylated mRNA is continues to be degraded from the 3′ end by the RNA exosome complex to degrade the RNA body resulting in the generation of a residual cap structure. The cap structure is subsequently degraded by the scavenger-decapping enzyme, DcpS which hydrolyzes the cap structure to releases m⁷GMP.

A. Dcp2 decapping enzyme

Dcp2 was the first decapping enzyme identified and is highly conserved in eukaryotes. It is a member of the Nudix family of proteins with a central domain consisting of a Nudix fold structure, which catalyzes the decapping step. Biochemical analyses have shown that Dcp2 can hydrolyze both monomethyl (m⁷G) and trimethyl (m^{2, 2, 7}G) capped RNA with poor activity on unmethylated capped (G-cap) RNA [13, 26–28]. Dcp2 is an RNA binding protein and requires a capped RNA substrate that is longer than 25 nucleotides and the RNA body contributes to the substrate specificity for Dcp2 decapping [28, 29]. Structural analyses of Dcp2 revealed it directly interacts with the cap and RNA body to recognize its substrate where the C-terminus of the NUDIX domain forms a conserved RNA binding channel to accommodate the RNA [30]. The RNA-binding property of Dcp2 is consistent with the observed transcript specificity. Dcp2 preferentially binds a 5′ terminal stem loop structure termed Dcp2 binding and decapping element (DBDE), which promotes recruitment of Dcp2 and subsequent decapping [31]. The DBDE is not restricted to a specific primary sequence and consists of at least an 8 basepair long stem with an intervening loop positioned within the 5′terminal 10 nucleotides of an mRNA for optimal Dcp2-mediated decapping.

Dcp2 is not ubiquitously expressed in all tissues and is developmentally regulated. Robust levels of Dcp2 protein are detected in the mouse embryonic brain, heart, liver and kidney, while it can only be detected in the corresponding adult brain but not the latter three tissues [16]. The selective function of Dcp2 in transcript decapping is further substantiated by genome-wide profiling of Dcp2 responsive mRNAs, which identified approximately 200 mRNAs that increased upon reduction of Dcp2 protein levels. Interestingly, a disproportional number of these mRNAs are involved in innate immunity and reduced Dcp2 levels correlate with enhanced resistance to viral challenge [32]. These findings demonstrate that Dcp2 is not a default decapping enzyme that functions on all mRNAs.

B. Nudt16 decapping protein

The transcript specificity of Dcp2 indicated that additional decapping enzymes are present in mammalian cells. Consistent with this premise, a second decapping protein Nudt16, was also shown to possess mRNA decapping activity [16]. Similar to Dcp2, Nudt16 is a member of the Nudix family of hydrolases. It was initially identified as a 29 kD nuclear protein in Xenopus (X29) that selectively bound and decapped the U8 snoRNA in vitro [33]. It was subsequently shown to be conserved across metazoans and thought to be a nuclear snoRNA decapping protein [34]. More recently, human Nudt16 was shown to be localized to both the nuclear and cytoplasmic compartments [16] [35] and regulate the stability of a subset of mRNAs in cells including the Angiomotin-like 2 mRNA [16]. Intriguingly, in addition to regulating the stability of distinct mRNAs, Dcp2 and Nudt16 coordinately function on select mRNA. Both enzymes are involved in the decay of the c-myc mRNA where the half-life incrementally increases with reduction of each protein, but most significantly increase when both are simultaneously reduced [15]. Conversely, the two decapping proteins function redundantly on the c-jun mRNA. A reduction of either decapping protein had no adverse consequence on stability of the c-jun mRNA while an increased half-life was only detected upon reduction of both [15].

Dcp2 and Nudt16 each influencing the stability of a small subset of mRNAs suggests there are as yet additional mRNA decapping enzymes that remain unidentified. Mammalian genomes contain 22 Nudt family proteins, two of which are Dcp2 and Nudt16. Whether any of the remaining 20 Nudt proteins also function in mRNA decapping remains to be determined. Interestingly, yeast contains 6 putative Nudix proteins, one of which is Dcp2. Although a homolog of Nudt16 is not evident in S. cerevisiae, whether the remaining 5 Nudix proteins possess intrinsic mRNA decapping activity is not yet known.

IV. Presence of an aberrant cap decapping protein in Saccharomyces cerevisiae.

The selective decapping of m⁷G-capped mRNAs by Dcp2 and Nudt16 and the requirement of a 5′ monophosphate substrate RNA for Rat1 and Xrn1 raises an interesting question regarding the decay of mRNAs containing either an unmethylated cap or lacking a cap altogether. These later two mRNAs with an aberrant 5′ end would be precluded from Dcp2 or Nudt16 decapping and would be protected from 5′-end exonucleolytic decay. The recent identification of the Rat1 interacting protein, Rai1 as a protein with both pyrophosphohydrolase and decapping endonuclease activities[36, 37] revealed the existence of a novel quality control mechanism that initiates the decay of aberrantly capped mRNAs.

A. Rai1 is a pyrophosphohydrolase that removes the 5′-terminal diphosphate from an uncapped RNA.

Rai1 was initially identified as an uncharacterized polypeptide in highly purified preparations of Rat1 [25] and subsequently shown to be a potent stimulator of Rat1 exonuclease activity [38]. Based on the intrinsic unstable nature of Rat1 in vitro [25] and the lack of detectable biochemical activity, Rai1 was proposed to stimulate Rat1 activity by stabilization of the Rat1 protein. Structural analysis of the Rat1-Rai1 heterodimer, as well as Rai1 alone revealed a conserved octahedral-coordinated sphere of a cation within Rai1 and the putative mammalian homology, Dom3Z [37]. Biochemical analysis demonstrated Rai1 possesses pyrophosphohydrolytic activity, which specifically cleaves within the 5′ triphosphate to release the terminal two phosphates and generates a monophosphorylated 5′ end RNA (pppRNA → pp + pRNA) that can subsequently serve as substrate for Rat1 [37]. Importantly, a mutually stimulatory association exists between Rat1 and Rai1. In addition to the stimulation of the Rat1 exonuclease activity by Rai1, Rat1 in turn stimulates the pyrophospholytic activity of Rai1 [37] thus comprising a mutually complementary degradation complex that can detect and degrade mRNAs lacking a 5′ end cap (Fig 2).

Fig 2. Rai1 possesses RNA 5’ end phosphohydrolase activity.

In yeast cells, Rai1 cleaves the triphosphorylated 5′-end of an mRNA within the α and β phosphates to generate phyrophosphate and release a 5′-end monophosphorylated RNA that is subsequently degraded by 5′-to-3′ exoribonuclease Rat1. Rai1 and Rat1 form a stable heterodimer and mutually enhance each other’s respective activities.

B. Identification of Rai1 as a decapping endonuclease protein that selectively functions on aberrantly capped mRNAs.

The ability of Rai1 to specifically cleave a triphosphate linkage between the α and β phosphates and generate a 5′-end monophosphorylated mRNA is reminiscent of the Dcp2 and Nudt16 mRNA decapping enzymes. This similarity indicates Rai1 may also function on capped RNAs. Addressing this question yielded two surprising findings. First, Rai1 can function on capped RNAs, but preferentially functions on mRNAs possessing an unmethylated cap [36]. Second, the mode of hydrolysis shifts from a pyrophosphohydrolase activity within the triphosphate linkage to a phosphodiesterase decapping endonuclease activity exclusively following the penultimate nucleotide to remove the entire cap structure (GpppN-RNA → GpppN + pRNA) [36] (Fig 3). Importantly, despite the altered site of cleavage, the 5′ end of the resulting RNA contains a monophosphate and can be degraded by Rat1. This unexpected activity of Rai1 in vitro was also evident in cells harboring the temperature sensitive ABD1 N7 methyltransferase mutant allele (abd1-5) which generates capped but not methylated primary transcripts at the nonpermissive temperature [39]. Steady state levels and stability of mRNAs increased in cap methylation deficient strains containing a mutant Rai1 gene (abd1-5 rai1Δ), implicating Rai1 in the regulation of aberrantly capped mRNAs lacking an N7-methyl moiety [36]. The combined activities of Rai1 in hydrolyzing an uncapped RNA retaining a 5′-end triphosphate or an RNA containing an unmethylated cap strongly implicates Rai1 in a quality control mechanism that initiates the demise of mRNAs containing an aberrant 5′ end. Moreover, the presence of such an activity directed against aberrant, but not normal, caps further supports the growing body of evidence that like other steps of pre-mRNA processing, capping is also a regulated step.

Fig 3. Rai1 possesses decapping endonuclease.

Rai1 possesses a decapping endonuclease activity that removes the entire cap structure from an mRNA containing a 5′cap lacking the N7-methyl moiety. As in Figure 2, the generated 5′-end monophosphorylated RNA is degraded by Rat1.

C. Rai1 stimulates Rat1 activity

Rai1 forms within a heterodimeric complex with Rat1 and stimulates its exonucleolytic activity [25, 37, 38]. Conversely, both the pyrophosphohydrolase and decapping endonuclease activities of Rai1 are stimulated by Rat1 [36, 37], demonstrating the significance of the heterodimer formation for their respective functions. Co-crystal structure of the Rat1-Rai1 heterodimer reveals an interaction interface involving the β8-αE segment and β4 strand of Rai1 [37] (please see Chapters 7 and 8 for structural parameters of Rat1 and Rai1). Mutations within the protein-protein interaction interface ablate the stimulatory activities of the two proteins without altering their respective basal level of catalytic activity [36, 37]. Through the formation of a Rat1-Rai1 heterodimer the complex functions as a molecular motor that detects and degrades mRNAs with aberrant 5′ ends regardless if it is uncapped or capped without the N7 methyl group. Rat1 will stimulate the hydrolase activities of Rai1 to generate a 5′-end monophosphorylated RNA that is degraded by the Rat1 subunit, which is in turn stimulated by Rai1 (Fig 2 and 3).

Rai1 appears to contribute to the enhanced activity of Rat1 by stabilization of the Rat1 structure [37]. The Rat1-Rai1 heterodimer is more stable in vitro [38] compared to Rat1 alone [25] and the C-terminal domain of Rat1 is essential for its activity [40]. The interaction of Rai1 with a loop contained within the critical C-terminal domain of Rat1 that is on the opposite side of the Rat1 active site, is proposed to stabilize the C-terminal loop structure which in turn indirectly coordinates and stabilize the Rat1 active site [37].

D. Generation of Aberrantly Capped mRNAs

Upon exposure to environmental stress, mRNAs are primarily released from polysomes and sequestered into structures termed stress granules as an adaptive response in eukaryotic cells [41–43]. Stress granules are cytoplasmic mRNP structures that contain an array of RNA-binding proteins, translation initiation factors, large and small ribosomal subunit protein components and mRNAs (for reviews, see [44–46]. Following return to an unstressed state, stress granules dissociate and the silenced mRNAs return to the polysomal pool of translating mRNAs [45]. Although a concomitant decrease in overall transcription occurs following stress, a complete transcriptional inhibition does not proceed and a diverse array of genes are either up- or down-regulated [47–49]. Such a situation presents the cell with an apparent paradox, at a time when cytoplasmic mRNAs are being sequestered, the nucleus is transcribing mRNAs to be transported to the cytoplasm only to be sequestered. Yeast cells have evolved an intriguing mechanism to address this apparent dilemma by regulating the extent to which the nascent transcripts generated upon exposure to stress are aberrantly capped. mRNAs that are either uncapped or incompletely capped by a guanosine lacking the N7 methyl moiety are subjected to 5′ end hydrolysis by the Rai1 protein to generate a 5′ monophosphorylated mRNA that would be degraded by the Rat1 protein [36].

E. mRNA Capping Under Nutrient Stress

Exposure of S. cerevisiae to either glucose or amino acid starvation leads to rapid decay of two relatively stable mRNAs, PGK1 and ACT1, in cells grown under normal growth conditions [36]. It should be noted that the increased instability is observed only when the studies were carried out with the incorporation of a lag phase to enable clearing of preexisting mRNAs generated prior to the onset of the stress condition. Importantly, a similar instability was not detected in an isogenic strain disrupted for the Rai1 gene where the mRNAs remained significantly more stable [36]. These findings were interpreted to indicate that addition of the m⁷G cap is inefficient when cells are deprived of nutrients and Rai1 selectively hydrolyzes their 5′ end to expose a 5′ monophosphorylated mRNA for Rat1-directed decay (Fig 2 and 3). Immunoprecipitation of capped mRNAs under conditions that separate m⁷G capped mRNA from aberrantly unmethylated capped or uncapped mRNAs [36] confirmed the aberrantly capped nature of the newly synthesized mRNAs under the stress parameters. The aberrant capped RNA decapping activity of Rai1 and 5′ to 3′ exoribonuclease activity of Rat1, indicate that Rai1-Rat1 heterodimer plays an essential role in clearing mRNAs with aberrant 5′-end caps. Rai1-Rat1 appear to be involved in a quality control mechanism that ensures mRNA 5′-end integrity by an aberrant-cap mediated mRNA decay mechanism (Fig 4).

Fig 4. Schematic diagram of Rat1-Rai1 heterodimer functions on mRNA 5’ end capping quality control.

RNA polymerase II (Pol II) nascent transcripts are initially capped at the 5′end with an N7 methyl cap structure (m7GpppN) co-transcriptionally. The pre-mRNA is further processed by splicing (not shown) and polyadenylation and the resulting mature mRNA is transported from the nucleus to the cytoplasm to undergo mRNA translation and/or decay. Transcripts lacking an m⁷G cap are targeted by the Rai1 phosphohydrolase or decapping endonuclease activity to generate a 5′-end monophosphated mRNA that is cleared by the exoribonuclease activity of Rat1.

V. Additional potential functions of Rai1

A. Rai1 in transcription termination and 5′ end mRNA capping quality control

Addition of the 5′ cap is cotranscriptional and occurs at the initial stages of transcription following synthesis of the first 20–40nt when both the capping and methyltransferase enzymes are associated with Ser5 phosphorylated C-Terminal Domain (CTD) of RNAP II [50, 51]. This corresponds to the checkpoint pause stage when capping is believed to occur and RNAP II enters an elongation phase upon Ser2 phosphorylation and subsequently Ser7 phosphorylation [51–56]. Several lines of recent evidence indicates that this coupling may promote the synthesis of only properly capped mRNAs and has a built-in quality control mechanism to clear aberrantly capped mRNAs. Disruption of the Ceg1 capping enzyme gene or the Abd1 N7-methyltransferase gene results in nonviable yeast cells demonstrating that the capping and methylation steps are both essential functions. Interestingly, disruption of the normal timing of transcription yields aberrantly capped pre-mRNAs. An RNA polymerase II with a catalytic site mutation that leads to a reduction in the rate of transcription (rpb1-N488D; [57]), generates mRNAs lacking a cap [58]. Moreover, similar to observations in ceg1 mutants harboring a temperature sensitive Rat1 disruption allele that results in increased mRNA levels [59], mRNA levels also increase in rpb1-N488D-rat1-1 cells with a corresponding increase in uncapped mRNAs [58]. The decrease in capping was attributed to a reduction of promoter occupancy of Ceg1 protein due to less efficient CTD Ser-5 phosphorylation. These findings demonstrate the importance of Rat1 in the cotranscriptional clearing of uncapped pre-mRNAs in a process perhaps analogous to the “torpedo” model of termination where Rat1 facilitates transcription termination following cleavage/polyadenylation at the 3′ end of a gene [60–62]. However, one important distinction is that the downstream RNA fragment following cleavage at the polyadenylation site contains a 5′ monophosphate that is an effective substrate for Rat1. In contrast, the 5′ end of an uncapped primary transcript would be expected to contain a 5′ triphosphate, which would be resistant to the Rat1 5′-end monophosphate RNA-specific exonuclease [25, 63]. Although not yet directly demonstrated, it is highly likely that Rai1 is also present early in the transcription cycle and cleaves either the triphosphate or unmethylated capped 5′ end to enable Rat1 access to the pre-mRNA.

C. Rai1 in rRNA processing

Rat1 has long been shown to be involved in the proper 5′-end processing of 5.8S and 25S rRNAs [54–56]. The 5.8S rRNA consists of two species, 5.8S_s and 5.8S_l, that differ in their 5′ length. The 5.8S_s is the predominant form and is generated by an endonucleolytic cleavage at site A3 followed by exonucleolytic trimming by Rat1 [38, 64]. Yeast strains disrupted for Rai1 are compromised in their production of 5.8S_s and predominantly shift to generating the 5.8S_l form, which is reversed with Rat1 overexpression [38] indicating the significance of Rai1 is to stimulate Rat1 activity. Surprisingly, an apparent Rat1-independent function of Rai1 was also noted in 5.8S rRNA processing where defects in 3′ end processing were detected [38]. Importantly, the defect in 5.8S 3′ end processing was analogous to defects observed in strains disrupted of the nuclear 3′ to 5′ exonuclease, Rrp6 suggestive of a functional interplay between Rai1 and Rrp6 [65] in addition to the interaction with Rat1. At least in yeast 5.8S rRNA maturation, Rai1 influences both Rat1 and Rrp6 exonucleolytic activity. Consistent with the absence of the Rat1 interaction domain within the putative mammalian homolog of Rai1, Dom3Z [37], Dom3Z could not be detected to interact with Xrn2 by two hybrid analyses [66]. However, yeast two hybrid interactions are detected between Dom3Z and the mammalian homolog of Rrp6, PM/Scl-100 [67] suggesting Dom3Z may influence 3′ end decay in mammals.

VI. Future Direction

A. Mammalian Rai1 Homolog

Rai1 shares weak primary sequence identity with a putative mammalian homolog, Dom3Z [38] but significant structural identity with Dom3Z [37]. The conserved structure of Dom3Z and Rai1 at the active site indicates Dom3Z may also possess catalytic activity. Dom3Z is a nuclear protein that can also be localized to a subset of cytoplasmic P bodies detected by GFP-Dcp1a [68]. Furthermore, knockdown of Dom3Z in mammalian U2OS cells, results in a marked reduction of cytoplasmic P bodies [68]. These observations suggest Dom3Z may have a role in mRNA metabolism in both the nucleus and cytoplasm. Future studies on the role of Dom3Z in potential aberrant cap regulation and mRNA metabolism are necessary to begin delineating the potential role of Dom3Z in mRNA catabolism.

B. Regulation of Capping

The overwhelming perception has generally been that addition of the 5′-end cap is an unregulated process that always proceeds to completion. The demonstration that capping is inefficient upon exposure of yeast cells to nutrient starvation indicates this perception may be more an indication of our naiveté and opens a new realm of regulation previously unforeseen. It is likely that cells exploit, and accentuate, an already existing regulatory mechanism following exposure to nutrient stress. Whether additional stress conditions also elicit the generation of aberrantly capped mRNAs remains to be determined although early indications are that at least heat shock also leads to aberrantly capped mRNA in yeast (X.J. and M.K., unpublished observations). A more pressing question is whether capping is normally inefficient and fulfills a modulatory process to maintain functional mRNA homeostasis. Further studies will begin address these pressing questions in addition to the mechanism by which any mode of regulation is controlled and the factors involved.