Infection by phytopathogenic bacteria triggers massive changes in plant gene expression, which are thought to be mostly a result of transcriptional reprogramming. However, evidence is accumulating that plants additionally use post-transcriptional regulation of immune-responsive mRNAs as a strategic weapon to shape the defense-related transcriptome. Cellular RNA-binding proteins regulate RNA stability, splicing or mRNA export of immune-response transcripts. In particular, mutants defective in alternative splicing of resistance genes exhibit compromised disease resistance. Furthermore, detection of bacterial pathogens induces the differential expression of small non-coding RNAs including microRNAs that impact the host defense transcriptome. Phytopathogenic bacteria in turn have evolved effector proteins to inhibit biogenesis and/or activity of cellular microRNAs. Whereas RNA silencing has long been known as an antiviral defense response, recent findings also reveal a major role of this process in antibacterial defense. Here we review the function of RNA-binding proteins and small RNA-directed post-transcriptional regulation in antibacterial defense. We mainly focus on studies that used the model system Arabidopsis thaliana and also discuss selected examples from other plants.
Invading phytopathogenic microbes encounter an elaborate defense system termed the plant immune system (Jones & Dangl, 2006). Host-encoded surface receptors recognize pathogen- or microbe-associated molecular patterns (PAMPs or MAMPs), such as the highly conserved N-terminal region of the bacterial flagellin or its synthetic derivative flg22 that is sensed by the Leucine-rich repeat (LRR) receptor-like kinase FLAGELLIN SENSING 2 (FLS2). This recognition elicits the PAMP-triggered immunity (PTI) that is widely active against microbes (Boller & He, 2009). Some pathogenic bacteria, in turn, inject effector proteins into the host cell that antagonize PTI components and allow successful infection. This, however, is a double-edged strategy, as some effectors referred to as avirulence (Avr) proteins, are recognized by intracellular plant resistance (R) proteins which mostly belong to the nucleotide binding site (NB)-LRR class. This interaction results in another layer of defense, the effector-triggered immunity (ETI) or R-gene-mediated resistance. ETI elicits similar responses as PTI albeit with a higher level and a faster kinetics and restricts growth of specific pathogens (Tao et al., 2003; Navarro et al., 2004; Tsuda & Katagiri, 2010).
Both PTI and ETI trigger massive reprogramming of the transcriptome (Tao et al., 2003; Navarro et al., 2004). The regulation of immune-response genes has been mostly studied at the transcriptional level. For example, a key component involved in the transcriptional regulation of immune-response genes is NONEXPRESSOR OF PATHOGENESIS RELATED GENES 1 (NPR1) that recruits transcription factors to defense-related genes including the PATHOGENESIS-RELATED 1 (PR1) gene that is responsive to the phytohormone salicylic acid (SA) (Dong, 2004). Much less is known about post-transcriptional events shaping the defense-related transcriptome.
Throughout their life, mRNAs dynamically associate with RNA-binding proteins (RBPs) that define pre-mRNA processing, lifetime, export from the nucleus, and translation (Glisovic et al., 2008). In Arabidopsis, RBPs have been functionally characterized in development, abiotic stress responses, flowering time or circadian rhythms (Staiger, 2001; Lorkovic, 2009). Several RBPs also have been implicated in pathogen defense (Qi et al., 2010). Additionally, endogenous small RNAs have emerged as key components orchestrating antibacterial defense (Ruiz-Ferrer & Voinnet, 2009; Katiyar-Agarwal & Jin, 2010). In this review, we discuss recent insights into post-transcriptional processes in antibacterial defense involving both, RBPs and small RNAs (Fig. 1), with a particular focus on Arabidopsis.
II. RNA decay
1. Regulated RNA stability
Cells modulate the stability of mRNAs in response to endogenous or environmental signals to rapidly adjust steady-state abundance of mRNAs. These alterations in mRNA decay are mediated by cis-acting sequence motifs and cognate RBPs. In Arabidopsis it was estimated that 1% of the transcripts have a half-life of < 60 min, and that these highly unstable transcripts are often associated with the circadian timing system and the response to mechanical stimulation (Gutierrez et al., 2002). With respect to pathogen defense, early on destabilization of a transcript encoding the PvPRP1 (Phaseolus vulgaris proline-rich protein (1) cell wall protein in bean has been reported in response to the elicitor of Colletotrichum lindemuthianum (Zhang et al., 1993). The PvPRP1 3′ untranslated region (UTR) interacts with a 50 kDa protein that presumably has a role in mRNA destabilization, as its RNA-binding activity inversely correlates with PvPRP1 mRNA levels (Zhang & Mehdy, 1994).
2. Nonsense-mediated decay (NMD)
A specialized pathway of RNA decay is NMD that removes mRNAs with premature termination codons (PTCs) that might give rise to aberrant proteins (McGlincy & Smith, 2008). Apart from this quality control, NMD more globally modulates the levels of a plethora of transcripts, for example, in sugar signaling (Yoine et al., 2006). In Arabidopsis, homologs of the NMD components UPF1 (UP FRAMESHIFT 1), UPF2, UPF3 and SMG7 have been recently characterized in antibacterial defense (Jeong et al., 2012; Rayson et al., 2012; Riehs-Kearnan et al., 2012; Shi et al., 2012). Transcript profiling for genes misexpressed in upf1-5 and upf3-1 mutants uncovered an overrepresentation of genes connected to pathogen response. The SA-responsive genes PR1, PR2 and PR5 are constitutively expressed in the upf mutants and also reach a higher level upon bacterial infection relative to wild type plants. Furthermore, SA content, levels of SA signalling components, and basal resistance to the virulent Pseudomonas syringae pv tomato strain DC3000 (Pto DC3000) are enhanced in the upf mutants. UPF1 and UPF3 mRNAs are themselves down-regulated in response to Pto DC3000 (Jeong et al., 2012), suggesting that down-regulation of NMD is part of an antibacterial defense response, potentially acting to stabilize PTC-containing transcripts encoding, for example, truncated R gene products (cf. III.2.). The constitutive SA-dependent defense and enhanced basal resistance observed in the NMD mutants could also be a consequence of the constitutive activation of R proteins, as recently described in the acd11 (accelerated cell death 11) and mpk4 (mitogen activated protein kinase 4) mutants (Palma et al., 2010; Kong et al., 2012; Zhang et al., 2012). In this case, R proteins would ‘guard’ UPF proteins, that is, detect modifications of UPF proteins that are targeted by pathogen effectors (Van der Biezen & Jones, 1998). Identifying primary mRNA targets of UPF proteins in genetic backgrounds where R proteins are destabilized and in bacterial challenged conditions will reveal the extent to which defense-related transcripts are directly controlled by NMD in the context of antibacterial defense.
III. Pre-mRNA splicing
1. Splicing components
Introns within pre-mRNAs are removed by splicing through the spliceosome machinery which contains small nuclear RNAs (snRNAs) and numerous RBPs (Reddy et al., 2011). Combinatorial usage of different splice sites within a pre-mRNA generates alternative splice (AS) forms. AS increases proteome complexity by generating protein variants with different domain composition and impacts gene expression in response to abiotic stress including high and low temperatures or heavy metals. The differential use of AS sites is controlled predominantly by serine/arginine-rich (SR) proteins and heterogenous nuclear ribonucleoproteins (hnRNPs) (Reddy et al., 2011).
A mutant with constitutive defense responses has been particularly instrumental to unravel the importance of splicing components in pathogen defense. This snc1 (SUPPRESSOR OF NPR1-1, CONSTITUTIVE1) mutant restores pathogen resistance in the npr1 mutant that is impaired in SA-dependent resistance (Zhang et al., 2003). SNC1 encodes an NB-LRR type R-protein. The snc1 mutation leads to constitutive resistance in the absence of pathogens by activating both, NPR1-dependent and NPR1-independent defense responses (Zhang et al., 2003). Suppressors of the snc1 phenotype have been identified and several of these MODIFIER OF snc1 (MOS) proteins are associated with RNA processing (Palma et al., 2007).
One of them, MOS4, shows homology to human BREAST CANCER-AMPLIFIED SEQUENCE2 and forms a complex with the AtCDC5 (CELL DIVISION CYCLE 5) Myb-transcription factor and PRL1 (PLEIOTROPIC REULATORY LOCUS 1), a WD-40 repeat protein also involved in sugar signaling (Németh et al., 1998; Palma et al., 2007). The complex has been termed the MOS4-associated complex (MAC), and MOS4, AtCDC5 and PRL1 are required for basal and R-gene-mediated defense (Palma et al., 2007).
Intriguingly, the homologs of these three proteins in humans and yeast are also part of a protein complex (Koncz et al., 2012). This complex is designated the NineTeen complex (NTC) in yeast or the PRP19 complex in humans, for its component Precursor RNA Processing 19, a U-box containing E3 ligase that binds both the E2 ubiquitin-conjugating enzyme and the substrates. The NTC has been implicated in spliceosome assembly and activation, and it was suggested that the MAC serves a similar function. Thus, affinity-purification of epitope-tagged MOS4 was performed to identify additional components (Monaghan et al., 2009). This approach identified two closely related proteins, MAC3A and MAC3B, with sequence homology to Prp19. Mac3a mac3b double mutants are compromised in both basal and R-mediated defense. MAC3A localizes to the nucleus and interacts with AtCDC5, suggesting that it is indeed an additional MAC component.
2. Alternative splicing
AS of R genes has been shown to play a role in pathogen defense. Initially, AS of the tobacco N gene that confers resistance to tobacco mosaic virus (TMV) was shown to give rise to a shorter NS transcript encoding functional N protein and a longer NL transcript encoding a truncated protein (Dinesh-Kumar & Baker, 2000). A switch from NS to NL and the corresponding protein isoforms during infection appears to be required for full resistance.
In Arabidopsis, the R protein RPS4 (RESISTANCE TO PSEUDOMONAS SYRINGAE 4) activates defense responses to avirulent pathogens expressing AvrRps4. RPS4 exists in several AS forms encoding truncated protein variants and AS is required for its function (Zhang & Gassmann, 2003, 2007). Upon inoculation with Pseudomonas syringae expressing AvrRps4 an AS form retaining intron3 strongly increases. Due to a PTC, this AS form is predicted to encode a truncated protein containing only the first four LRRs. Plants which cannot produce this AS form are susceptible to Pto DC3000 (AvrRps4). Because other AS forms are also important for resistance, individual RPS4 splice isoforms likely fulfill separate functions. PTC-containing transcripts potentially can be translated into truncated proteins or even peptides that can compete with the full-length protein. In analogy with siRNAs and miRNAs they have been designated small interfering peptides or micro proteins (Seo et al., 2012). It is conceivable that PTC-containing R protein isoforms may interfere with functions of the authentic protein in plant immunity. This could potentially enhance the pathogen effector recognition repertoire that has to cope with rapidly evolving effector proteins.
The AS pattern of RPS4 is altered in the mos12-1 mutant (Xu et al., 2012). Furthermore, retention of the second and third intron in the SNC1 transcript was favoured in mos12-1 with a concomitant reduction in SNC1 protein abundance. Accordingly, MOS12 is required for basal and RPS4-mediated immunity. MOS12 shows homology to cyclin L in humans and harbours an atypical SR-rich domain that may interact with other splicing factors. MOS12 localized to the nucleus and associated with the MAC in vivo. These findings led the authors to analyze SNC1 and RPS4 AS patterns in other mac mutants. Indeed they found changes in splicing patterns of SNC1 in mos4, Atcdc5 and mac3a mac3b, providing compelling evidence that MAC mediates AS of a subset of R genes. The mos6-1 mutation may also be linked to splicing, as MOS6/importinα-3 is required for nuclear import of PRL1 (Németh et al., 1998; Palma et al., 2005; Koncz et al., 2012).
Another mutant, mos14, also shows impaired AS of SNC1 and RPS4 and pathogen susceptibility (Xu et al., 2011). MOS14 encodes a protein with homology to transportin-SR that mediates nuclear import of SR proteins. Indeed, MOS14 interacts with AtRAN1, a RAS-related GTP-binding protein likely involved in nuclear protein translocation, and with SR proteins, suggesting that it might facilitate the nuclear import of SR proteins involved in AS of SNC1 and RPS4. Overall, these studies show that dynamic accumulation of AS forms is a recurrent theme in fine-tuning R gene function in different plant species.
AS is often linked to NMD of PTC-containing AS isoforms, resulting in changes of transcript levels (McGlincy & Smith, 2008). The hnRNP-like RBP AtGRP7 (Arabidopsis thaliana GLYCINE-RICH RNA-BINDING PROTEIN7) negatively auto-regulates by AS generating a PTC-containing isoform that decays via NMD (Staiger et al., 2003; Schöning et al., 2008). AtGRP7 is part of the circadian timing system and plays a role in PTI and resistance against Pto DC3000 (Staiger & Heintzen, 1999; Fu et al., 2007). AtGRP7 is the first RBP shown to be targeted by a Pto DC3000 effector protein, the ADP ribosyltransferase HopU1 (Fu et al., 2007). HopU1 ADP-ribosylates the conserved arginine 49 residue within the RRM (RNA recognition motif) of AtGRP7 that is crucial for its RNA-binding activity and in vivo function (Schöning et al., 2007; Jeong et al., 2011). Thus, HopU1 likely interferes with the processing of defense-related transcripts by interfering with AtGRP7′s activity (Fig. 2). In line with this, PR transcript levels are elevated in plants over-expressing AtGRP7 (Streitner et al., 2010). Furthermore, AtGRP7 affects AS of multiple transcripts by direct binding in vivo (Streitner et al., 2012). AtGRP7 shuttles between nucleus and cytoplasm which points to functions in RNA processing both in the nucleus and the cytoplasm (Kim et al., 2008; Lummer et al., 2011). HopU1 additionally targets chloroplast-localized RBPs but their function in immunity has not yet been established (Fu et al., 2007). Overall, the immunity-related phenotypes of mutants defective in splicing components and the requirement for R gene AS in disease resistance emphasize a crucial role for splicing in plant immunity.
IV. The nuclear mRNA export machinery
Export of mRNAs across the nuclear envelope for translation in the cytoplasm proceeds through nuclear pore complexes (NPCs). NPCs consist of subcomplexes of nucleoporin proteins (Nups) surrounding a central channel. MRNA export is initiated by binding of RBPs that make contact to export receptors which in turn contact the NPC (Chinnusamy et al., 2008).
MOS3, also identified in the snc1 suppressor screen, localized to the nuclear envelope and is the orthologue of vertebrate Nup96 (Zhang & Li, 2005; Parry et al., 2006). Nup96 is part of the Nup107-160 subcomplex required for mRNA export. Because Nup96 is necessary for immunity in mice, Wiermer and colleagues screened T-DNA mutants in all Arabidopsis orthologues of the Nup170-160 constituents for their pathogen response (Wiermer et al., 2012). Only mutants defective in Nup160 and Seh1 were impaired in basal and R-gene mediated resistance and showed aberrant poly(A) RNA accumulation in the nucleus. Poly(A) RNA export was also impaired in the hrp1-4 mutant that shows enhanced susceptibility to Pto DC3000 (Pan et al., 2012). HRP1 encodes a nuclear protein related to human HPR1, a component of the conserved THO/transcription export complex required for mRNA export in humans and yeast. Further experiments are required to understand whether the phenotypes are caused by nuclear retention of specific defense-associated mRNAs or a more global decrease in translation.
V. General RBPs
A protein with three RRMs, AtRBP-DR1 (Arabidopsis thaliana RBP-defense response 1) was originally identified as an interactor of RPS2 (Qi et al., 2010). Because an Atrbp-dr1 mutant is more sensitive to Pto DC3000 but not to avirulent strains, AtRBP-DR1 was implicated in basal defense but not R-gene-mediated resistance. Constitutive over-expression of AtRBP-DR1 led to increased SA production and up-regulation of PR1. As AtRBP-DR1 localizes to both the nucleus and the cytoplasm, it may affect any step of processing of defense-related transcripts.
The mos2-1 mutant has a lesion in a protein with a G-patch domain and two KOW motifs, both predicted to bind RNA (Zhang et al., 2005). MOS2 is involved in basal defense and R-gene-mediated resistance. It is located in the nucleus, and a potential connection of MOS2 to the spliceosome has been pointed out (Koncz et al., 2012).
Another protein co-purifying with MOS4, MAC5A, contains an RRM and a zinc finger motif and is implicated in pathogen defense, as the mac5a-1 mutant partially suppresses the snc1 phenotype (Monaghan et al., 2010). The human counterpart of MAC5A is RBM22 that interacts with the spliceosomal U6 RNA and pre-mRNA and takes part in the splicing mechanism (Koncz et al., 2012; Rasche et al., 2012). This suggests that MAC5A may also be involved in pre-mRNA splicing in Arabidopsis.
VI. Gene silencing
1. Small RNAs biogenesis
A large fraction of the transcriptome comprises small noncoding RNAs that regulate various steps of gene expression. Based on their biogenesis, they are classified into microRNAs (miRNAs) or small interfering RNAs (siRNAs). MiRNAs are generated from endogenous transcripts with internal stem-loop structures and regulate multiple developmental processes, the transition to flowering and stress responses (Voinnet, 2009). In Arabidopsis, these imperfectly base-paired pri-miRNAs are converted into pre (precursor)-miRNAs by DICER-LIKE1 (DCL1), one of the four DCL proteins. The pre-miRNAs are processed to miRNA/miRNA* duplexes. 2′-O-methylation by the methyltransferase HUA ENHANCER1 contributes to stabilization of miRNAs by preventing their uridylation and degradation (Ren et al., 2012; Zhao et al., 2012). One strand of the miRNA/miRNA* duplex is assembled into RISCs (RNA-induced silencing complexes). Most miRNAs associate with ARGONAUTE1 (AGO1), one of the 10 AGO family members. MiRNAs direct AGO1-RISC to mRNA targets by virtue of perfect or partial sequence complementarity. The miRNA-mRNA interaction leads to mRNA cleavage executed by the endonucleolytic ‘slicer’ activity of AGO1, or to translational inhibition (Brodersen et al., 2008). In animals, miRNA-mediated translational repression is followed by removal of the poly(A) tail and mRNA decay but these temporal molecular events have not yet been systematically investigated in plants (Bazzini et al., 2012; Djuranovic et al., 2012). Apart from post-transcriptional control, miRNAs can affect transcription of target loci by directing DNA methylation (Bao et al., 2004; Wu et al., 2010).
Endogenous siRNAs are generated from extensive double-stranded RNAs (dsRNAs) including natural antisense transcripts (NATs), inverted repeats and even miRNA precursors or originate through the action of RNA-dependent RNA polymerases (RDRs) (Chellappan et al., 2010; Katiyar-Agarwal & Jin, 2010). The dsRNAs are processed into 21, 22 and 24 nt-long siRNAs through DCL4, DCL2 and DCL3, respectively, giving rise to heterogenous populations of small RNAs. SiRNAs are also loaded into AGO-containing RISCs that are recruited to their targets. The largest class of siRNAs is represented by c. 24 nt long heterochromatic siRNAs, mainly derived from repeats and transposons. They act in a process termed RNA-directed DNA methylation to cause epigenetic modification (Law & Jacobsen, 2010). This nuclear silencing pathway has recently been characterized in antibacterial and antiviral defense, but will not be discussed in this review (Raja et al., 2008; Lopez et al., 2011; Dowen et al., 2012).
In addition to the biogenesis of primary siRNAs, plants and other organisms have evolved the production of secondary siRNAs as a feed-forward amplification of silencing signals. In plants, these secondary siRNAs are produced by the combined action of primary siRNA/miRNA-directed transcript cleavage and the activity of RdRPs that use the target transcripts as template to generate dsRNAs. In plants, the best-characterized secondary siRNAs are termed trans-acting siRNAs (tasiRNAs). TasiRNAs are produced from miRNA-directed cleavage of non-coding TAS primary transcripts by the action of RDR6 that transcribes the 3′ cleavage product, leading to dsRNAs that are sequentially processed by DCL4 into phased siRNAs (Allen et al., 2005). These siRNAs in turn can target protein-coding transcripts in trans. Amplification of siRNAs occurs also from miRNA-targeted protein-coding transcripts through a small RNA biogenesis that resembles tasiRNAs. It was recently shown that a unique class of miRNAs of 22 nts trigger the production of secondary siRNAs (Chen et al., 2010; Cuperus et al., 2010). Importantly, secondary siRNAs and tasiRNAs have the potential to control gene expression in a non-cell autonomous manner and therefore might play a significant role in short-distance and long-distance immune responses.
2. MicroRNAs in plant immunity
MiRNAs were implicated in plant immunity by the observation that the dcl1-9 mutant defective in miRNA biogenesis and ago1-25 and ago1-27 mutants defective in miRNA activity are impaired in PTI (Navarro et al., 2008; Li et al., 2010). The first miRNA directly shown to be involved in plant immunity was miR393. MiR393 levels increased in response to flg22 and the Pto DC3000 hrcC mutant that cannot inject effector proteins into the host cell (Navarro et al., 2006, 2008; Fahlgren et al., 2007). The miR393 target, the F-box auxin receptor TIR1 (TRANSPORT INHIBITOR RESPONSE 1), targets Aux/IAA (Auxin/indole-3-acetic acid) proteins for degradation. The Aux/IAAs heterodimerize with AUXIN RESPONSE FACTOR (ARF) transcription factors, thereby preventing the transcription of auxin-responsive genes. Thus, flg22 induction leads to stabilization of IAA proteins (Fig. 3). This suggested that miRNA393-mediated suppression of auxin signaling is part of the plant defense response, although a miR393-independent effect, likely involving transcriptional repression of auxin receptors, was also reported. The role of miR393 in antibacterial resistance was further demonstrated by showing that over-expression of MIR393 restricted growth of Pto DC3000, whereas over-expression of the auxin receptor ABF1, which is partly refractory to miR393-directed cleavage, led to enhanced susceptibility (Navarro et al., 2006; Robert-Seilaniantz et al., 2011).
Arabidopsis contains two genes, MIR393a and MIR393b that are processed into an identical mature miR393. The miRNA* strand was initially considered as a nonfunctional by-product of miRNA biogenesis (Jones-Rhoades et al., 2006). A recent study, however, implicates the miR393b* strand, which has one mismatch to miR393a*, in plant immunity. MiR393b* was identified in the small RNA population bound to AGO2, the only AGO induced during antibacterial defense (Zhang et al., 2011). This miRNA* was shown to target MEMB12 encoding a Golgi SNARE protein implicated in vesicle trafficking and protein secretion. A loss-of-function mutant, memb12-1, was more resistant to Pto DC3000 and Pto DC3000 (AvrRpt2). Notably, more antimicrobial PR1 protein was secreted into the intercellular fluid of memb12-1 infected with Pto DC3000 (AvrRpt2) compared to wild type plants, consistent with disturbance of intracellular vesicle trafficking. Thus, the miRNA guide strand and its miRNA* partner both are involved in different aspects of plant immunity via different AGO-RISCs (Fig. 4). The miRNA strand with a uridine residue at its 5′ terminus associates with AGO1 that has a known preference for U (Mi et al., 2008; Montgomery et al., 2008), while the miRNA* strand bearing an adenine at its 5′ end associates with AGO2 that prefers A. Thus, only in one case the choice of the miRNA guide strand occurs according to the prediction that the strand with the 5′ terminus located at the thermodynamically less-stable end of the miR393/miR393* duplex serves as guide strand.
MiR393 was also identified among miRNAs elevated upon treatment with the Pto DC3000 hrcC mutant, along with miR160 and miR167, which target ARFs. This implies a more general role of miRNA-mediated suppression of auxin-signaling in immune responses (Fahlgren et al., 2007). Notably, among miRNAs with reduced levels after challenge with the Pto DC3000 hrcC mutant were miR168, which targets AGO1, and miR162, which targets DCL1 (Fahlgren et al., 2007), suggesting that the biogenesis and activity of miRNAs might be stimulated during antibacterial defense. This is consistent with the critical role of the miRNA pathway in PTI. Interestingly, several other miRNAs were down-regulated during antimicrobial defense, some of which target mRNAs encoding for canonical disease resistance proteins (Klevebring et al., 2009; Shivaprasad et al., 2012). This highlights a complex regulatory network of induced miRNAs, which negatively regulate repressors of plant defense, and repressed miRNAs that negatively regulate positive regulators of plant defense.
3. Endogenous siRNAs in plant immunity
Several reports indicate a key role of endogenous siRNAs in antibacterial resistance. The first example of a NAT-derived siRNA associated with plant immunity was nat-siRNAATGB2 (Katiyar-Agarwal et al., 2006). It was proposed to originate from the overlap of a NAT pair encoding a Rab2-like GTP-binding protein, ATGB2, and a mitochondrial pentatricopeptide repeat protein-like protein, PPRL, that appears to act as a negative regulator of RPS2-mediated resistance. AtGB2 is up-regulated in response to Pto (AvrRpt2), and siRNAATGB2 originating from the dsRNA at the overlap of AtGB2 and PPRL has been suggested to cause down-regulation of PPRL (Katiyar-Agarwal et al., 2006). A recent study using genome-wide direct sequencing of mRNA 3′ends showed that the PPRL mRNAs does not extend to the region complementary to the siRNA, raising the question whether another mechanism may contribute to the reciprocal expression of AtGB2 and PPRL (Sherstnev et al., 2012).
A representative of so-called long siRNAs (lsiRNAs) is AtlsiRNA-1 that comprises 39-41 nt-long RNAs and is proposed to derive from a NAT pair consisting of a transcript encoding a receptor-like kinase and the antisense transcript AtRAP (Katiyar-Agarwal et al., 2007). The receptor-like kinase transcript is induced in response to Pto DC3000 (AvrRpt2) with concomitant production of AtlsiRNA-1 and downregulation of AtRAP. AtRAP is a negative regulator of both, PTI and ETI. In mammalian cells, the RAP domain is enriched in proteins bound to mRNA in vivo (Baltz et al., 2012). This suggests that AtRAP may indeed have RNA binding activity and contribute to regulation at the RNA level.
Endogenous siRNAs have also emerged as regulators of mRNAs encoding for disease resistance proteins. The Columbia RPP4 (RECOGNITION OF PERONOSPORA PARASITICA 5) locus, named RPP5 in Landsberg erecta, was the first disease resistance locus shown to be controlled by post-transcriptional gene silencing. This locus is composed of seven NB-LRR genes, which arose from local duplications and rearrangements, and includes SNC1. The detection of convergent, overlapping sense and antisense transcripts along with 21-nt small RNAs corresponding to several R genes at the RPP4 cluster, suggested an auto-regulation of these R genes by post-transcriptional gene silencing (Yi & Richards, 2007). Consistent with this hypothesis, higher accumulation of SNC1 transcripts were detected in ago1 and dcl4 loss of function mutants. Based on these findings, the authors proposed that siRNAs targeting the RPP4 locus might act as a sensor to detect any interference in RNA silencing activity triggered by pathogens. In this scenario, pathogen-triggered release of silencing would enhance expression of R genes leading to defense activation. The auto-regulation at the RPP4 locus by RNA silencing is also likely to be necessary to maintain a low basal expression level of R genes in normal growth conditions, likewise preventing their potential threats to plant fitness.
4. MiRNA-directed siRNA biogenesis in plant immunity
MiRNA-triggered production of secondary siRNAs has recently emerged as a key regulatory process that ensures a tight control of disease resistance gene expression and therefore extended the discoveries made at the RPP4 locus. For example, the tobacco N gene was shown to be targeted by two miRNAs. One of them, nta-miR6019, is 22 nt long and triggers the production of RDR6- and DCL4-dependent siRNAs (Li et al., 2012). Interestingly, overexpression of N together with nta-miR6019/6020 in N. benthamiana impaired N-mediated resistance to TMV, suggesting that these miRNAs and secondary siRNAs have a functional role in N-regulation. This mechanism is widespread across Solanaceae as many mRNAs encoding for disease resistance genes were also shown to be targeted by 22 nt miRNAs and secondary siRNAs. Another study conducted in Solanum lycopersicum provided further biological relevance for such a process in the context of pathogen infection (Shivaprasad et al., 2012). The authors studied the miR482/2118 superfamily, which has the potential to control the expression of a large set of NB-LRR genes in different plant species. They showed that miR482 targets multiple NB-LRR genes leading to the production of RDR6-dependent secondary siRNAs. Some of them control in turn secondary targets in trans, including a transcript encoding for a protein that has homology to PAD3, which is involved in camalexin biosynthesis. Importantly, miR482 levels decreased in response to Pto DC3000 and Tobacco rattle virus, which both encode silencing suppressors (see VII. Perspective). Therefore, pathogen-triggered suppression of RNA silencing likely derepresses a whole repertoire of NB-LRR transcripts, which might enhance pathogen effector recognition and/or contribute to basal defense responses. Interestingly, miR482 is related to the Arabidopsis 22 nt long microRNA miR472, which also targets NBS-LRRs and initiates the production of secondary siRNAs (Chen et al., 2010; Cuperus et al., 2010). However, the dynamics and biological relevance of these small RNAs in Arabidopsis innate immunity remains to be established.
RNA-based regulation has emerged as a critical layer of control in plant immunity. Forward and reverse genetic screens for components involved in PTI or ETI in Arabidopsis have identified proteins associated with RNA-binding activity, splicing, and mRNA export. To uncover the molecular underpinnings of their action, target transcripts associated with these proteins in vivo have to be disclosed in the future by RNA immunoprecipitation. Further, an intricate interconnection of post-transcriptional control with transcriptional regulation is envisaged. Splicing and transcription are interconnected through the NTC. As many components of the MAC, the Arabidopsis NTC equivalent, are involved in pathogen defense, their influence may also extend to transcriptional control (Perales & Bentley, 2009). For example, in mammalian cells comprehensive cataloguing of the mRNA-bound proteome identified many proteins involved in transcriptional regulation (Baltz et al., 2012). Moreover, differential recruitment of transcripts to polyribosomes upon powdery mildew infection in barley and virus infection in Arabidopsis suggests that regulation of translation initiation shapes the defense proteome (Moeller et al., 2012). This offers the added advantage of being readily reversible when the pathogen is gone. Together, the regulatory mechanisms discussed here imply that one has to be aware of the limitations when profiling changes in gene expression in response to pathogens based on RNA steady-state abundance only.
There is now compelling evidence that plant small RNAs play a key role in controlling the innate immune response. So far, the few functionally characterized small RNAs in innate immunity represent probably the tip of the iceberg and it is likely that many additional small RNAs will be soon implicated in this biological process. The recent identification of hundreds of non-conserved miRNAs raises the intriguing possibility that a significant proportion of those may have arisen to control the expression of rapidly evolving protein-coding genes involved in innate immunity. Identifying the extent to which these ‘young’ miRNAs indeed control the innate immune response, and understanding their transcriptional regulation, remains to be further investigated. This subset of non-conserved miRNAs, including the ones that regulate disease resistance genes, might also shed novel lights on the mechanisms by which miRNAs are generated and selected during evolution.
RNA silencing has long been appreciated as an antiviral defense response. However, recent findings also implicate the miRNA pathway as a major component of antibacterial basal resistance. As a corollary, several effectors from phytopathogenic bacteria have evolved to inhibit miRNA biogenesis and activity (Navarro et al., 2008), although the mechanisms involved remain to be further elucidated. It will be particularly important to identify these bacterial suppressors of silencing that can directly interact with, and alter the function of, host RNA silencing factors. Because such silencing suppressors likely target key residues that are essential for the normal function of RNA silencing factors, we are anticipating that those bacterial proteins will soon be considered as valuable molecular tools to identify novel RNA silencing factors and unravel their function and regulation in both RNA silencing and innate immunity.
Generally, the view of small RNAs as regulators of mRNA expression has been extended to a more global view of reciprocal control between miRNAs and their targets (Salmena et al., 2011; Pasquinelli, 2012). Any RNA with a miRNA target site can potentially function as a competing RNA. The dual function of mRNAs in protein coding and regulation of miRNA function likely will influence the interpretation of pathogen-induced miRNA regulation of their targets (Fig. 5a): For example, mRNAs binding to miRNAs may prevent their function, and alterations in the mRNA complement upon pathogen attack could then release the miRNA. The efficient targeting of miRNAs can also be influenced by RBPs. RBPs could modulate miRISC access to overlapping target sites by sterical hindrance (Fig. 5b). Alternatively, binding of RBPs could induce secondary structure changes that allow or prevent access of RISCs (Fig. 5c).
Moreover, components are shared between silencing pathways and RNA metabolism. For example, DCL4 is also involved in transcription termination of an endogenous Arabidopsis gene, FCA, encoding an RBP involved in flowering time control (Liu et al., 2012). Thus, proteins involved in silencing may additionally influence the defense transcriptome via direct effects on host transcripts. In conclusion, extensive cross-talk between the RBP networks and miRNA networks is envisaged, making it a challenging future task to unravel the targets of post-transcriptional control in plant immunity.
We thank Dr Jiradet Gloggnitzer for insightful comments on the manuscript, and Dr Armin Hallmann for help with the figures. Work in DS's Lab is supported by the DFG (STA 653/3; STA 653/4-1) and in LN's Lab is supported by the ATIP/Avenir-Fondation Bettencourt Schueller Programme ‘Emergences’ (Mairie de Paris).