Processing bodies and plant development

Introduction

Development is a dynamic process resulting from spatiotemporally controlled gene expression. An increasing body of evidence suggests that post-transcriptional regulation plays important roles in regulating gene expression during development.

The unstable and mobile nature of RNA facilitates both sequence-dependent and independent regulation of gene expression. Cytoplasmic RNA decay serves as the final step to regulate gene expression post-transcriptionally. The 5′ to 3′ degradation of RNAs is a major pathway for cytoplasmic RNA decay, which involves multiple steps including deadenylation, decapping and exonuclease digestion. Among those steps, decapping, the removal of 5′ m7GDP is a rate-limiting step which takes place in P-bodies [1]. Other RNA decay pathways including the 3′ to 5′ decay have been reviewed elsewhere [2–4]. Here, we will focus on recent research relating to P-bodies and their role in plant development.

P-bodies which appear as cytoplasmic foci in eukaryotes cells [5,6] contain RNA-protein particles (RNPs) assembled from RNA and protein components [7]. The evolutionary conservation of P-body proteins in eukaryotes allowed us to identify several Arabidopsis thaliana P-body components by sequence-based genomic surveys.

Protein components of plant P-body

Table 1 shows known and putative P-body components in Arabidopsis homologous to characterized P-body proteins in yeast and human. Plant P-bodies contain VCS, a human Hedls homolog that is not found in yeast, indicating plant P-bodies are structurally closer to human P-bodies than those in yeast. Nevertheless, Arabidopsis DCP1 and DCP2 which mark plant P-bodies well can also localize to yeast P-bodies [8••,9•,10••]. It is not yet clear whether yeast, plant and mammalian P-body structures are truly homologous or are just merely similar.

Amongst P-body components investigated, DCP2 is the only one possessing decapping enzymatic activity in vitro [8••]. DCP5 alone has been shown to be a translation suppressor in wheat germ extracts [11••]. Co-localization of DCP1, DCP2, VCS, DCP5, XRN4 and DHH1 suggests plant P-bodies function in both translational repression and mRNA decapping [8••,9•,10••,11••,12•]. We prefer a kinetic model that DCP1, DCP5 and possibly DHH1 are involved in forming RNPs for translational repression of target mRNAs. The untranslated mRNAs then are subjected to decapping by recruiting DCP2 and VCS and the decapped RNAs are subsequently digested by XRN4 (Figure 1) [11••].

The Arabidopsis genome also encodes small RNA-mediated decay pathway proteins. Although AGO1 was reported to localize in nuclear dicing bodies [13], the cytoplasmic fraction of AGO1 was found residing in P-bodies [14•,15••]. There is as yet no Arabidopsis homolog of GW182, which connects AGO proteins to mammalian P-bodies. A possible candidate to functionally supplant GW182 is SDE3, which co-localizes with DCP1 in plants (Xu and Chua, unpublished). SDE3 is a MOV10 homolog in plants but contains a GW motif known to mediate interactions between GW182 and AGOs [16]. UPF1 which is a nonsense-mediated decay (NMD) pathway factor also localizes to plant P-bodies (Xu and Chua, unpublished). Although eIF4E marks P-bodies in human and yeast cells [17] it does not localize to plant P-bodies [8••], instead it localizes to plant stress granules (SGs) [10••]. Plant P-bodies share many conserved proteins with yeast and human P-bodies indicating that plant P-bodies execute similar functions, i.e. translational repression and decapping. However, plant P-bodies also contain their own distinct protein components which may form functionally diverged RNPs.

Besides the known homologs (Table 1), other proteins have also been found in plant P-bodies. For example, ISE2 granules represent plant-specific SGs and a small portion of ISE2 has been shown to co-localize with DCP2 [18]. AtTZF1 and hTTP have been reported to co-localize with plant AGO1, DCP2 and XRN4 in P-bodies [15••] and CMV 2b, a viral gene-silencing suppressor, is also associated with P-bodies [14•]. More plant-specific P-body components are expected to emerge with further advances in plant functional genomic studies.

Function implications by phenotypic analysis

Severe single-mutant alleles of DCP1, DCP2, DCP5 and VCS are lethal at the seeding stage with pleiotropic phenotypes including vein, leaf blade and root defects, suggesting the affected proteins are required for post-embryonic development [8••,9•,11••,12•]. The developmental phenotype of mutants defective in P-body component is rather unexpected for three reasons. (1) Among the P-body components only DCP2 has decapping activity in vitro [8••]. The requirement for other components for growth suggests that DCP2 activity in vivo is rate-limiting. The complex formation between DCP2, DCP1, DCP5 and VCS is either indispensible to the mRNA decapping step or DCP1, DCP5 and VCS, may act in steps upstream of decapping. One possibility is that translationally repressed RNP formation is necessary and precedes decapping. Consistent with this notion, DCP5 alone is able to suppress mRNA translation in vitro [11••]. (2) Accumulation of capped mRNAs in decapping-defective mutants results in severe seedling lethality, indicating that these mRNAs are not subject to alternate mechanisms of RNA decay and are able to re-engage with ribosomes for translation. Under this condition, the accumulation of both mRNAs and their products has the same effect as the over-expression of cognate genes. (3) It is not likely that the deadenylation step precedes the decapping step, or the majority of these mRNA substrates are subject to co-translational decapping which may bypass the deadenylation step [39].

Amongst mutants deficient in P-body components the most severe phenotype is found with lsm4; in this mutant, embryos do not develop beyond the globular stage (Table 1). upf1-3 is seedling lethal similar to moderate dcp2 mutants. The developmental defects in mutants of P-body components are pleiotropic, suggesting multi-target effects are common for the affected proteins. Moreover, developmental lethality in P-body mutants suggests that P-body components are required for developmental transition or reprogramming. For example, dcp2-1 seeds are able to germinate but seedling growth ceases with the typical lethal phenotype appearing 6 d after imbibition [8••]. It is likely that P-bodies are required to degrade seed-stored mRNAs, over-accumulation of which would otherwise perturb developmental transition.

RNA substrates of plant P-body

P-bodies contain RNPs assembled from mRNAs and proteins. Although the assembly mechanism of individual RNPs is largely unknown, knowledge of mRNA substrates provides important clues to understanding functions of RNPs or P-bodies in general. As biochemical purification of P-bodies has so far been elusive for unknown reasons, analysis of genetic alleles of specific P-body component has become an important strategy to identify and characterize RNA compositions of plant P-bodies.

Seed storage proteins (SSP) mRNAs are P-body substrates in dried seed and during germination. In dcp1-1, dcp1-2 and dcp5-1, SSP mRNAs accumulate greatly at 6 d after germination [11••,12•]. Similarly, SSP protein levels are increased in these mutants, suggesting that P-bodies target these mRNAs for both translational repression and decapping. Dried seeds accumulate high levels of DCP1 and DCP5 but not DCP2, which is consistent with the storage of SSP mRNAs in seeds without degradation. After germination, DCP1 and DCP5 levels decline along with increase in DCP2 levels and decapping and degradation of SSP mRNAs. In dried seed, SSP mRNAs are stored and translationally arrested but become degraded during germination. These observations are in line with the kinetic model suggesting that translation repression and RNP formation precedes the decapping step. Decapping, therefore, is crucial to the developmental transition from a quiescent state and to rapid post-embryonic growth. Besides SSP mRNAs, many unstable mRNAs [40] including EXPL1 and SEN1 are also stabilized in decapping mutants indicating that a broad spectrum of mRNAs are subjected to translational repression and decapping [8••,11••].

EBF1 and EBF2 mRNAs are likely the direct targets of EIN5/XRN4. In ein5 mutant, EBF1 and EBF2 mRNAs are highly expressed, resulting in reduced levels of EIN3 [22,23]. Although its enzyme activity has not yet been directly demonstrated XRN4 is presumably an exoribonuclease preferring RNA substrates with 5′-monophosphate.

Assuming this activity of XRN4, the increased levels of EBF1/2 mRNAs in xrn4 mutant are likely to be uncapped. It is not clear how uncapped mRNAs are able to re-engage with ribosomes for translation. The other characterized targets of XRN4 includes 3′ products of miRNA cleavage [21], transcripts of the STM-GR transgene [20•] and also 130 endogenous transcripts determined by microarray analysis [24]. Another microarray experiment on uncapped RNAs demonstrates the presence of diverse uncapped RNAs in flowers [46]. Further work is needed to understand how uncapped mRNAs are channeled to cap-insensitive translation or are directly degraded or serve as substrates for dsRNA production.

Several lines of evidence support the view that P-body formation is likely not required for degradation of mRNAs targeted by miRNA, siRNA and ta-siRNA. (1) Although dcp5-1 mutant displays reduced formation of P-bodies it does not significantly accumulate miRNA targets [11••]. (2) In contrast to mammalian systems, the majority of miRNAs in plants show complete or near complete sequence complementarity to their mRNA targets, and 3′ cleavage products are usually readily detectable. (3) RNA interference has been successfully used in plants to achieve degradation/silencing of specific mRNAs in a sequence-dependent manner [41]. Notably, VCS has been found to be required for miRNA-guided translational repression but had no effect on mRNA levels of genes like SPL3, SCL6-IV and CIP4 [27]. It is not clear if the translational-arrested miRNA targets are stored in P-bodies and if P-bodies are required for this miRNA-guided translational repression. Levels of several miRNAs are reduced in decapping mutants suggesting a possible connection between miRNA biogenesis and plant P-bodies (Xu and Chua, unpublished).

Non-sense mediated decay (NMD) is a eukaryotic mRNA quality-control mechanism for elimination of aberrant mRNAs containing a premature termination codon (PTC) to prevent the production of truncated proteins. Although mRNA targets responsible for the developmental defect of upf 1 and upf 3 mutants have not been fully identified, several aberrant mRNAs containing PTC have been found to over-accumulate, i.e. SMD-7 and ATGRP7 transcripts [31,42]. A genome-wide survey revealed that expression of not only protein-coding transcripts but also many non-coding RNAs (ncRNAs) is upregulated in both upf 1 and upf 3 mutants [32••]. Typical ncRNAs include natural antisense transcripts, and transcripts of transposable elements and pseudogenes, all of which contain no or short ORFs and share the same features with canonical PTC-containing NMD substrates. Interestingly, UPF1, which is required for degradation of aberrant RNAs, appears to play an important role in RNA interference [30•].

As P-bodies contain many RNPs functional in regulating plant development it is not surprising that they may be involved in pathogen infection and defense. In yeast, brome mosaic virus (BMV) is localized to P-bodies and viral RNA-dependent RNA polymerase binds to the yeast lsm1p [43•]. More important, decapping complex factors including Lsm1p-7p/Pat1p/Dhh1p have been reported to be necessary for BMV replication in yeast [44•]. These results obtained with a heterozygous host suggest that BMV may likely replicate in plant P-bodies as well.

P-body formation

Franks and Lykke-Anderson proposed a kinetic model for P-body formation in which the formation of translationally repressed RNPs precedes the mRNA decapping step [45]. DCP5 is a translation repressor and does not stimulate DCP2 activity in vitro. However, P-body formation is reduced in the dcp5-1 mutant [10••], suggesting DCP5 is required for P-body formation. DCP1, DCP5 and possibly DHH1 are involved in forming RNPs for translational repression of target mRNAs. The untranslated mRNAs then are subjected to decapping by recruiting DCP2 and VCS and the uncapped products are subsequently degraded by XRN4 (Figure 1).

There is evidence that P-body formation is dispensable to active decapping, since DCP1 marked P-bodies are still detected in plants lacking DCP2 [11••]. P-bodies devoid of DCP2 most likely contain translationally repressed RNPs. The multi-steps between translation and mRNA decapping facilitate regulation of the decapping step. During stress signaling-pathways are found to regulate RNP formation through modifying DCP1 (Xu and Chua, unpublished). Therefore, translatability of a specific group of mRNAs is regulated by responses to environmental clues before being committed to the irreversible and rate-limiting step of decapping.

Conclusions

Plant P-bodies share many protein components with yeast and mammalian P-bodies, but are more closely related to mammalian P-bodies. However, plant P-bodies also contain specific components. The phenotype of mutants and specific substrates found in P-bodies suggest that their function are developmental rather than housekeeping. The most conserved functions of P-bodies are translational repression and mRNA decapping. P-bodies are more likely a compartment rather than a single RNP complex. Diversity and specificity uncovered from the characterization of many P-body-related RNA substrates indicate that translationally repressed RNPs from different RNA decay pathways may localize to this compartment. A major future challenge will be to understand how this diversity and specificity are achieved and genetic dissection using Arabidopsis to discover more P-body components and RNPs will certainly be an important approach.