Alternative Splicing in Plant Immunity

Alternative splicing (AS) occurs widely in plants and can provide the main source of transcriptome and proteome diversity in an organism. AS functions in a range of physiological processes, including plant disease resistance, but its biological roles and functional mechanisms remain poorly understood. Many plant disease resistance (R) genes undergo AS, and several R genes require alternatively spliced transcripts to produce R proteins that can specifically recognize pathogen invasion. In the finely-tuned process of R protein activation, the truncated isoforms generated by AS may participate in plant disease resistance either by suppressing the negative regulation of initiation of immunity, or by directly engaging in effector-triggered signaling. Although emerging research has shown the functional significance of AS in plant biotic stress responses, many aspects of this topic remain to be understood. Several interesting issues surrounding the AS of R genes, especially regarding its functional roles and regulation, will require innovative techniques and additional research to unravel.


Introduction
Alternative splicing (AS) describes the processing of a single pre-mRNA to produce multiple transcript isoforms [1]. Genome-wide studies have shown that AS is prevalent in eukaryotes and that more than 95% of human multi-exon genes undergo AS [2,3]. One of the most impressive examples of OPEN ACCESS be EDS1 (Enhanced disease susceptibility 1), a pivotal signal transducer in RPS6-mediated ETI, although EDS1 also functions downstream of pathogen detection [38].
PTI cannot completely inhibit pathogen colonization, but can retard pathogen invasion [39]. By contrast, ETI can be viewed as intensified and long-lasting PTI that includes the development of systemic acquired resistance and rapid, localized programmed cell death known as the hypersensitive response (HR) [40]. A chain of defense responses occur concomitant with the HR, including oxidative burst, accumulation of salicylic acid (SA), expression of pathogenesis-related (PR) genes, and defensin biosynthesis. PTI involves mitogen-activated protein kinase-signaling cascades and the accumulation of reactive oxygen species [41,42], and constitutive activation of PTI in the absence of pathogen results deleterious effects on plant development. As a long-lasting, systemic response, ETI must be fine-tuned to protect the plant from pathogen attack without excessive fitness costs.

R Genes
The majority of cloned R genes encode proteins containing a central nucleotide-binding site (NBS) and a C-terminal leucine-rich repeat (LRR) region. The NBS region normally consists of three subdomains, NBS, ARC1, and ARC2. The characteristic NBS subdomain includes a binding site for ATP or GTP and is active in initiation of signaling cascades leading to resistance responses [43]. The ARC subdomains (named for their presence in Apaf-1, R proteins, and CED-4) are highly conserved and essential for intramolecular interactions of R proteins [44]. By contrast, the LRR motif confers recognition specificity to the plant defense response [45][46][47][48].
Based on their N-terminal structures, members of the NBS-LRR family of R genes can be further subdivided into two subfamilies. One subfamily comprises members with a domain homologous to the intracellular signaling domains of the Drosophila Toll and mammalian Interleukin (IL)-1 receptor (TIR-NBS-LRR). TIR-NBS-LRR genes are exclusively present in dicot species. Members of this subfamily include tobacco N, flax L6 and M, Arabidopsis RPP1, RPP4 and RPS4, and Medicago truncatula RCT1. Another subfamily is characterized by a putative coiled-coil domain in the N-terminal region (CC-NBS-LRR). CC-NBS-LRR genes are widely distributed in both dicots and monocots. Both the CC and TIR domains likely function in interaction with downstream factors in ETI signaling [49]. Although most TIR-and CC-NBS-LRRs lack putative transmembrane domains or organelle-targeting signals and are predicted to be cytosolic, some show dynamic changes in subcellular localization [42,50].

Signaling Components in ETI
In addition to their structural differences, TIR-NBS-LRR and CC-NBS-LRR genes generally function through distinct signaling pathways, requiring either EDS1 or NDR1 (Non-race-specific disease resistance 1), respectively [51]. One exception is the Arabidopsis HRT gene that confers resistance to TCV (Turnip crinkle virus). HRT is a CC-NBS-LRR gene but its signaling is dependent on EDS1 [52]. Moreover, a few CC-NBS-LRR genes including RPP7, RPP8, and RPP13 can activate defense signaling independent of EDS1 and NDR1 [51,53,54]. Venugopal et al. [55] proposed, however, that EDS1 and SA act redundantly to regulate ETI to viral, bacterial, and oomycete pathogens. As such, participation of EDS1 in signaling triggered by CC-NBS-LRR R proteins may be masked by SA, and vice versa. In such cases, the requirement for EDS1 would be observed only when disease resistance does not require SA accumulation. PAD4 (Phytoalexin deficient 4) and SGA101 are indispensable for EDS1-required signaling to restrict pathogen growth [56,57]. EDS1, PAD4, SAG101 function independently, as well as in a ternary complex of SAG101-EDS1-PAD4, serving as signal transducers in HRT-mediated resistance to TCV [58]. However, the HR associated with TCV resistance conferred by HRT requires only EDS1, whereas the SA signaling induced by HRT requires only PAD4.
Genetic analysis of Arabidopsis mutants defective in systemic acquired resistance led to the isolation of NPR1 (Non-expresser of PR genes 1), which encodes a putative transcription factor regulating PR gene expression downstream of SA production [59]. Further investigation of the regulator of NPR1 in Arabidopsis resulted in identification of the gain-of-function mutant snc1 (Suppressor of npr1-1, constitutive 1) [60], which exhibits a dwarfed phenotype caused by constitutive activation of defense signaling in the absence of pathogen infection. Based on these mutants it can be concluded that wild type SNC1 suppresses NPR1 and to finely control autoimmune responses. Interestingly, snc1 encodes a TIR-NBS-LRR R protein, and the snc1 mutant morphology is restored or suppressed to different extents in a series of mos (Modifier of snc1) mutants. Thus far, 13 MOS genes have been cloned, the gene products of which act in various cellular and molecular processes, including pre-mRNA splicing, nuclear trafficking of serine-arginine rich (SR) proteins and protein modification, which is indicative of a highly complex network for regulation of R protein-mediated ETI [61][62][63][64][65][66][67][68][69][70][71][72].

AS of TIR-NBS-LRR Genes
Most TIR-NBS-LRR genes have conserved gene structures in the coding region, which generally contains three or four introns. The first exon encodes the TIR domain, the second exon encodes the NBS domain, and the remaining exons encode the LRR region. AS of TIR-NBS-LRR genes can result from intron retention, selection of alternative exons, or usage of alternative 5' or 3' splicing sites. Alternative isoforms have been reported for many TIR-NBS-LRR genes, such as tobacco N [73], flax L, and M loci [74], Arabidopsis SNC1, RPS4, RPS6, RPP5, and RAC1 [37,[75][76][77][78], tomato Bs4 [79], potato Y-1 [80], and M. truncatula RCT1 [81]. The functional consequences of AS events have been characterized for only a few TIR-NBS-LRR R genes, including Arabidopsis RPS4, tobacco N, and M. truncatula RCT1.

Arabidopsis RPS4
The Arabidopsis RPS4 gene confers resistance to Pseudomonas syringae pv. tomato strain DC3000 (DC3000) expressing AvrRps4. AS produces six transcript isoforms of RPS4 via retention of intron 2 and/or intron 3, and splicing of a cryptic intron in exon 3 ( Figure 1A) [28]. Due to premature stop codons introduced by frame shifts, the alternatively spliced isoforms encode no or fewer LRR repeats. Experiments involving stable transformation of RPS4 genomic constructs lacking intron 2 and/or intron 3, under the control of the RPS4 promoter, showed that deletion of a single intron was sufficient to abolish RPS4 function, even though splicing of remaining intron was unaffected and the normally spliced transcript was also expressed [28]. Therefore, resistance to DC3000 requires AS of RPS4. A role for these alternatively spliced isoforms as regulatory RNAs remains possible, but there is evidence that they encode truncated proteins that regulate the activity of full-length RPS4. An artificial combination of normal and alternatively spliced isoforms only partially restored RPS4-mediated resistance [28]. The molar ratio of RPS4 transcript isoforms in that experiment was altered compared to those naturally occurring, suggesting that the ratio is of functional importance. The abundance of the various AS isoforms of RPS4, particularly the isoform retaining intron 3 (RPS4 AT4 ), is under dynamic regulation in response to AvrRps4. Whereas the full-length transcript including all exons is the predominant splicing product in uninoculated leaves, pathogen inoculation induces a rapid, >100-fold increase of RPS4 AT4 [29]. The truncated proteins encoded by RPS4 variants were detected in transient expression assays, confirming that the aberrant transcripts are functional.

Tobacco N
Tobacco N specifically recognizes a 50-kDa helicase protein (p50) of tobacco mosaic virus (TMV), and the N gene is alternatively spliced [73,82]. In addition to the major isoform (N RT ), an alternative isoform (N AT ) is generated via AS of a hidden exon containing a stop codon within intron 3, which yields a putative product lacking 13 of 14 LRR repeats ( Figure 1B). Similar to RPS4, a dynamic abundance ratio of N RT to N AT is also observed during TMV infection [30]. Although N RT is predominant before infection, N AT is the more abundant isoform 6 h after TMV inoculation, and the original isoform ratio reappears 9 h after inoculation. Perturbing the ratio of N RT to N AT resulted in compromised TMV resistance. The boost in N AT production may result from a signaling cascade induced by interaction between N RT and p50. Because the accumulation of spliced variants occurs rapidly, the induced AS may regulate N function via feedback inhibition. Tobacco transformants expressing only N RT displayed incomplete resistance manifested by delayed HR, which suggests that N AT is required for full N-mediated resistance [30]. However, N AT expressed alone was not sufficient for TMV-dependent HR.

M. truncatula RCT1
RCT1 confers resistance against multiple races of Colletotrichum trifolii, a hemi-biotrophic fungal pathogen that causes anthracnose disease in Medicago [81]. AS of RCT1 results from the retention of intron 4, instead of intron 2 and/or intron 3 as in N and RPS4 ( Figure 1C). The alternative isoform (RCT1 AT ) is predicted to encode a truncated protein consisting of the entire TIR, NBS, and LRR domains, but lacking the C-terminal domain of the normal RCT1 protein (RCT1 RT ). RCT1-mediated resistance requires RCT1 AT and RCT1 RT , as transformants containing only RCT1 AT or RCT1 RT showed no anthracnose resistance [31]. Though the expression of RCT1 transcripts was stable and constitutive, and unaffected by pathogen infection, a certain expression threshold for RCT1 AT seemed to be essential for effective resistance.

Flax L6 and Tomato Bs4
In contrast to RPS4, N and RCT1, alternatively spliced transcripts of flax L6 and tomato BS4 are not required for full resistance to the corresponding pathogens. For example, transgenic plants carrying an intronless L6 (L6 RT ) exhibited complete rust resistance, similar to plants carrying the wild-type L6 ( Figure 1D) [74]. L6 triggers flax rust resistance by direct interaction with its cognate effector AvrL567. The flax rust resistance gene M, which is homologous to L, is also alternatively spliced; therefore, it is possible that AS of the M locus could functionally substitute for AS of L6. This hypothesis was supported by trans-complementation of the Rx gene. Rx is a CC-NBS-LRR gene that confers resistance to potato virus X (PVX) elicited by a coat protein. The CC domains in Rx share 96% similarity with those of Gpa2, which is required for resistance to potato nematode [83,84]. The function of an NBS-LRR derivative of Rx (Rx NBS-LRR) lacking the CC domain can be complemented by Gpa2 to induce coat protein-dependent HR [85]. Such trans-complementation appears to require high identity between the R proteins, because pepper Bs2 (Bacterial spot resistance gene 2), which is homologous to Rx, failed to functionally complement an Rx NBS-LRR derivative.
Transient co-expression of L6 RT and AvrL6 in tobacco gives rise to apparent HR, which argues against any interference by the M locus [86]. Likewise, transient expression of intronless Bs4 revealed that the normal Bs4 protein alone could mediate AvrBs4 recognition, which suggests that AS of Bs4 is functionally dispensable [79]. Whereas such transient expression assays have served well for isolation of R genes [83], whether this system can reliably be used to analyze functional roles for AS of R genes remains to be established. It is possible that the observed HR could be due to partial resistance conferred by an endogenous full-length R protein, such as tobacco N. Recent analysis of truncated R genes containing TIR-NBS only (TN) in Arabidopsis showed that chlorosis was induced by transient overexpression of TN genes [87]. The alternative L6 and Bs4 isoforms were not tested in transient assays; therefore, these transient expression experiments may not fully reflect the physiological roles of AS in the process.
A stunted phenotype caused by constitutive defense responses was observed in transgenic tobacco carrying an L6 genomic construct, as well as in transgenic tobacco plants in which L6 RT was under the control of the 35S promoter [88]. This evidence is suggestive that AS is irrelevant to L6-mediated resistance, with dwarfism serving as a reporter for activation of defense responses. However, the lack of tobacco transformants expressing L6 RT from its native promoter precludes firm conclusions about this. Structural and functional analysis demonstrated that the TIR domain alone is necessary and sufficient for L6 immune signaling [89]. More interestingly, with only one exception (L10-A), tobacco plants transformed with a genomic construct of L10 grew normally [88]. Further analysis revealed that the stunted phenotype of L10-A is associated with the presence of an additional truncated L10 transcript resulting from an aberrant T-DNA integration [88]. This truncated transcript is predicted to encode a protein containing the TIR and 39 amino acids of the NBS domain of L10. These findings point to the possibility that the functional significance of AS in L6 has been undervalued.

AS of CC-NBS-LRR Genes
AS has been identified in many CC-NBS-LRR R genes, including LR10 and Sr35 in wheat [90,91], Mla in barley [92,93], Pi-ta and RGA5 in rice [94,95], and JA1tr in common bean [96], but the functional importance of this post-transcriptional modification for full disease resistance is largely unknown. Only the alternative transcripts of RGA5 have been functionally characterized in a robust system [95].
Rice blast R protein RGA5 was found to cooperate with RGA4 in recognizing two sequence-unrelated effectors, Avr-pia and Avr1-CO39, through direct binding. Two transcript isoforms are generated by AS of the third of the three introns in the coding region of RGA5 [97]. As in the case of M. truncatula RCT1, protein products of both the intronless, fully-spliced transcript (RGA5 RT ) and the AS version (RGA5 AT ) share the CC, NBS, and LRR domains, and differ only in the C-terminal region, which is related to the copper binding protein ATX1 (RATX1) [95]. Transformants carrying RGA5 AT are fully susceptible to Avr-pia-and Avr1-CO39-expressing Magnaporthe oryzae strains. Furthermore, in conjunction with RGA4, RGA5 RT is necessary and sufficient to confer dual recognition specificity [95]. Yeast two-hybrid assays demonstrated that Avr-pia and Avr1-CO39 physically interact with the C-terminal RATX1 domain, which is present only in RGA5 RT . The disruption of the RATX1 domain consequently renders RGA5 AT inactive. These findings highlight the importance of the non-LRR regions near the C-termini of R proteins, indicating that they may deserve more attention when exploring the functions of R proteins in disease resistance.
Another rice blast resistance gene, Pi-ta, confers resistance to strains of M. oryzae containing cognate avirulence gene Avr-Pita. A total of 12 distinct transcript isoforms were identified as resulting from AS and are predicted to encode 11 proteins. Some of these transcripts are constitutively expressed while others show differential expression upon blast infection [94]. Their regulatory roles in disease resistance remain unknown.
The barley powdery mildew resistance genes Mla6 and Mla13 have very similar gene structures, including the conservation of two introns in the 5'-UTR and two introns in the coding region, as well as a large intron in the 3'-UTR. Notably, both genes exhibit AS of the 5'-UTR, which contains three upstream ORFs (uORFs); AS is also predicted to cause variation of one amino acid in the coding region of Mla13 [92]. The expression of Mla13 transcripts is induced upon pathogen penetration, and a dynamic change in the relative abundance of transcript isoforms has been observed. Inactivation of uORF translation via mutagenesis suggests the uORFs in the 5'-UTR downregulate Mla13 synthesis [98]. Hence, AS of uORFs may finely tune Mla13 expression to achieve effective resistance while minimizing host cell damage. However, it remains unknown whether full resistance mediated by Mla13 or Mla6 requires AS of the uORFs.

Possible Mechanisms of AS-Mediated Regulation of Defense Response
In the cases where AS is necessary for disease resistance, transgenic plants containing only the full-length transcript do not display auto-immunity or lesion mimic phenotypes induced by increased R protein activity, suggesting that AS is not likely to negatively regulate the R gene function. By contrast, the absence of AS impairs R gene-mediated resistance, which is indicative of positive roles for AS in defense responses. R protein isoforms therefore possibly function by suppressing the negative regulation of immunity activation, or by directly engaging in effector-trigged signaling, or by a combination of both.

Disruption of R Protein Autoinhibition
Whether an R protein is active or inactive is determined by the binding of ATP or ADP to the NBS domain [99]. Since constitutive activation of R proteins leads to lethal effects on plant growth, negative regulation of R protein activity is essential [100][101][102]. Intramolecular interactions between R-protein domains may function as a regulatory switch, and several mechanistic models have been proposed to describe this R protein self-regulation, such as the "Jack-knife" model [103]. These models are based largely on the trans-complementation of Rx CC-NBS and LRR domains [85]. From the crystal structures of the TIR and CC domains [89,104], Takken and Goverse proposed a model in which the NBS domain interacts with the N-terminal half of the LRRs, maintaining the R protein as inactive in a closed conformation before pathogen invasion [105]. An electrostatic interface that maintains the inactive conformation may be formed by interaction between the LRR and NBS domains. The C-terminal LRRs are exposed to serve as an antenna to detect charge changes induced by environmental perturbations. Since the TIR or CC domain can also interact with the NBS domain [85], the R protein is stabilized in a compact structure in the absence of pathogens. Studies on intramolecular interactions of Rx have provided evidence that the NBS domain alone is not sufficient for stable binding, but instead requires the CC domain. Notably, the CC domain could also interact with the NBS domain, unless N-terminal LRRs were bound to the NBS domain [44,85,106]. As such, the interaction of LRRs and NBS domains seems to cause conformational changes in the latter that facilitate NBS binding with CC domain. It has also been demonstrated that the ARC1 subdomain is necessary for binding of the Rx N-terminal LRR domain, while the ARC2 subdomain is required to maintain an autoinhibited state in the absence of elicitor, as well as for subsequent signaling [44]. Mutation in LRRs or conserved ACR2 motifs of the NBS domain leads to the autoactivation of Rx and RPS5 [44,85,107]. The majority of truncated R protein variants generated by AS are presumably unstable, due to the lack of LRR domain, and it is thus speculated that the aberrant R protein isoforms induced by pathogen inoculation could form intermolecular interactions with their regular protein products. This would disrupt the closed conformation stabilized by intermolecular interactions and free active R proteins.
In addition to the autoinhibition, R proteins are also subjected to negative regulation by trans factors [103]. RIN4, guarded by RPM1 and RPS2, is phosphorylated upon infection with P. syringae by AvrRpm1 and AvrB [108,109]. The rin4 mutants cannot survive in the presence of wild-type RPM1 and RPS2, due to strong activation of defense responses independent of pathogen infection. However, the rin4 defective phenotype is suppressed in the triple mutant rin4 rps2 rpm1 [103]. It was deduced that interactions of RIN4 with RPM1 and RPS2 negatively regulate the activities of both of these R proteins. The down-regulation of R protein activity could also be achieved by limiting its accumulation to a steady level. SRFR1 (Suppressor of RPS4-RLD1) interacts with SNC1 to negatively regulate production of several R proteins, such as RPS2, RPS4 and RPS6 [110]. Likewise, the F-box protein CPR1 (Constitutive expresser of PR genes 1) controls the stability of R proteins through SKP1-Cullin1-F-box (SCF)-mediated protein degradation [111]. Loss-of-function cpr1 mutants displayed higher expression of SNC1 and RPS2, as well as autoimmunity responses. Excess R protein isoforms produced via AS upon effector recognition may compete with full-length R protein to interact with negative regulators and decrease the relative abundance of these suppressors, thereby releasing active R protein [103]. This assumption is in line with observations that the overexpression of some R genes, including Rx, RPS2, and RPM1, leads to constitutive activation of resistance signaling.

Function as Signaling Factors
Overexpression of the TIR or CC domain of some R proteins (e.g., RPS4, RPP1, MLA10, and L6) can induce HR in the absence of cognate effectors [89,104,[112][113][114]. In addition to the TIR-NBS-LRR-encoding R genes, plants also contain short pseudo-R gene homologs (TN and TX) [115]. TN proteins contain the TIR and NBS domains, but lack the LRR domain, while TX proteins have only the TIR domain followed by a small and variable C-terminal domain. Arabidopsis contains 21 TN and 30 TX genes [116]. Transient and stable overexpression of some TN and TX genes induced necrosis in tobacco leaves and reduced disease symptoms in P. syringae-infected Arabidopsis plants, respectively [87]. This suggests that the truncated R proteins resulting from AS may also confer disease resistance with or without recognition specificity.
The crystal structures of the TIR domain of L6 and CC domain of MLA10 indicated that two activated R proteins form a homodimer at the CC or TIR domain to constitute a minimal functional unit [89,104]. In the presence of full-length R protein, the production of massive amounts of truncated proteins containing TIR or CC may serve as a rapid and energy-efficient mechanism to activate responses to pathogen infection. If so, the rapid increase of TIR or CC domain-dependent dimerization stimulated by AS of R genes might function to amplify the plant defense responses.
Protein function is associated with subcellular localization. It is possible that the alternative proteins generated by AS are localized to different compartments than the full-length R proteins, and numerous reports have demonstrated dynamic subcellular localization for R proteins such as RPS4 and N [42,117,118]. Distinct signaling pathways can be initiated by a single R protein in different subcellular localizations, and, thus, the coordinated trafficking of R proteins is required for the activation of full resistance [119]. RPS4 is detected in both the endomembrane and nucleus in healthy and diseased leaves, with RPS4 accumulation in the nucleus appearing to be necessary for AvrRPS4-trigged immunity [118]. AvrRPS4 also shows a nucleo-cytoplasmic distribution. Forcing AvrRPS4 to accumulate in cytoplasm through the C-terminal fusion of a nuclear export sequence led to moderate HR and partial suppression of bacterial growth. By contrast, sequestration of AvrRPS4 in the nucleus by fusion of nuclear localization sequence was sufficient for inhibition of bacterial growth, but cell death elicited by HR was abolished. HR signaling is therefore mediated by cytoplasmic RPS4-AvrRPS4 interaction, whereas the nuclear R-Avr interaction-induced resistance is not coupled to programmed cell death. This is in line with the findings that restriction of pathogen spread does not always correlate with HR [120][121][122]. However, because the construct used for examination of RPS4 subcellular localization consisted of its genomic sequence with an upstream fusion of the reporter gene under the control of the 35S promoter, any differential targeting of full-length RPS4 compared to truncated variants could not be distinguished [117]. It is likely that the truncated RPS4 proteins would accumulate in the endomembrane system, since their C termini lack a bipartite nuclear localization sequence, which is necessary for accumulation of full-length RPS4 in the nucleus. This could explain why only 6%-10% of RPS4 was observed in the nuclei. The distinct types of signaling triggered by nucleo-cytoplasmic distribution of R-Avr interaction may be coordinated by AS and differential localization of the resultant protein isoforms.
Tobacco N is predicted to be cytoplasmic because it does not carry a recognizable nuclear localization signal. Unexpectedly, it was found to be localized to both the cytoplasm and the nucleus, and nuclear localization is required for N function [117]. Different constructs, tagged with distinct fluorescence genes for different N transcript isoforms, are needed to test whether AS leads to diverse subcellular localizations for alternative N isoforms.

Regulation of AS of R Genes
AS dramatically increases the diversity of the transcriptome, and AS of R genes plays crucial roles in regulating plant defense responses; therefore, the mechanisms that regulate AS must be finely tuned to control the levels of different AS transcripts. Removal of introns within pre-mRNA in eukaryotes is catalyzed by the spliceosome, a highly dynamic and complex macromolecule comprising five (U1, U2, U4, U5, and U6) small ribonucleoproteins (snRNPs) and numerous RNA binding proteins (RBPs), such as serine/arginine-rich (SR) proteins and heterogeneous nuclear ribonucleoproteins (hnRNPs). The precise selection of intron/exons requires splicing factors to recognize four loosely conserved sequence features in pre-mRNA: (1) the 5' splicing site (SS) of GU paired with snRNP U1; (2) a branch point A for binding of splicing factor 1 at the 18 to 40 nucleotides upstream of the 3' SS; (3) the 3' SS of AG and (4) a poly-pyrimidine tract for recruitment of U2 auxiliary factor heterodimer [26,123]. It is noteworthy that a single intron may contain multiple sites for each of these four conserved sequence elements, adding more complexity in splicing site selection.
Differential selection of 5'-or 3'-SSs can be also affected by some short sequences of cis-elements in intronic and exonic region. According to the position and function, these cis-elements are grouped as exonic splicing enhancers, exonic splicing silencers, intronic splicing enhancers, and intronic splicing silencers. These splicing regulatory elements bind to trans-acting splicing factors, such as SR proteins and hnRNPs, playing critical roles in both constitutive and alternative splicing through either inducing or suppressing selection of nearby 5'-or 3'-splicing sites [124,125]. Interestingly, SR genes are also extensively alternatively spliced and AS of SR genes is affected by environmental stresses such as temperature, light and salt, which in turn induces splicing changes in the pre-mRNAs of other genes [126].
As mentioned above, screening for suppressors of the gain-of-function mutation snc1 led to the identification of a set of MOS genes, some of which function in pre-mRNA processing. For example, Arabidopsis mutants carrying a loss-of-function mutation for MOS4, MOS12, or MOS14 show altered splicing patterns for SNC1 and RPS4, which indicate that those genes have regulatory roles in AS of R genes [62,71,72]. MOS4, required for both ETI and PTI, is a nuclear localized CC homologous to human BCA2 (Breast cancer-amplified sequence 2). Together with the Myb-transcription factor CDC5L (Cell divison cycle 5 like protein) and the WD-40 repeat PLRG1 (Pleiotropic regulator 1), BCA2 was isolated from humans as an important component of a multiprotein spliceosome complex that includes the E3 ubiquitin ligase Prp19 (Precursor RNA processing 19) [127,128]. Yeast two-hybrid and in planta assays confirmed that MOS4 interacted with the Arabidopsis homologs of CDC5L and PRLG1 (AtCDC5 and PRL1, respectively) to constitute a core structure for a spliceosome-associated complex termed the MOS4-associated complex (MAC) [62]. MAC3A and MAC3B, two functionally redundant homologs of Prp19, contribute to proper splicing of SNC1, though their effects on AS of RPS4 have not been investigated [129]. Similarly, whether two other redundant homologs, MAC5A and MAC5B, function in R gene AS has not been tested [130]. However, given that its counterpart in human is RBM22, which interacts with U6 snRNP, it is possible that MAC5 participates in pre-mRNA splicing in plants.
MOS12 encodes an SR protein homologous to human cyclin L [72]. Co-immunoprecipitation of MOS12 with MOS4 indicates that MOS12 is also associated with the MAC. The mos12 mutant displays compromised RPS4-mediated resistance as well as an altered splicing pattern of RPS4, leading to a different abundance ratio of RPS4 transcript isoforms. However, the splicing pattern of RPS6 is normal in the mos12 mutant, as is RPS6-mediated resistance. This suggests that in addition to MAC, more spliceosomal complexes with distinct splicing specificities probably exist in plants.
Impaired SNC1-and RPS4-mediated PTI and ETI was also observed in the loss-of-function mutant of MOS14 [71]. In addition to distorted splicing patterns, the mos14 mutants showed reduced expression of SNC1 and RPS4. MOS14 encodes a nuclear protein homologous to transportin-SR, which functions in nuclear trafficking of the SR protein. MOS14 interacts with four different SR proteins through its C-terminus, while the N-terminus interacts with a GTP-binding protein AtRAN1 (Ras-related nuclear protein 1) which functions in many processes, including nuclear transport of proteins. The nuclear localization of these four proteins was disrupted in mos14 mutants, which consequently affects the splicing profiles for their targets. Defective splicing resulting from mislocalization of MOS14 cargos may cause the reduction in SNC1 and RPS4 expression [71].

Future Prospects
The functional importance of AS in plant disease resistance has become increasingly clear. However, despite the substantial progress that has been made in the past decade, AS research in plant immunity is still in its infancy. The AS events characterized to date in CC-NBS-LRR genes appear not to be required for disease resistance. Therefore, more research on CC-NBS-LRR genes will be needed to confirm whether AS plays other functional roles. Some truncated TIR-NBS-LRR proteins encoded by alternative transcripts are required for full R-gene mediated resistance, which raises an interesting question of how they engage in R-Avr interactions. Neither the "elicitor-receptor" nor the "guard hypothesis" models explain how truncated R proteins function in triggering plant defense responses. The cognate Avr proteins for L6, RPS4 and N, and even their host targets [117,119,131] have been identified, providing an opportunity to discover biological roles for the aberrant R protein variants. Research in this direction will likely provide insights into the functional mechanisms of truncated R proteins, as well as their dynamic subcellular localization, and interactions with Avr proteins and their targets, thus extending our understanding of gene-for-gene resistance in plants.
The alternative transcript isoforms may be subjected to nonsense-mediated decay (NMD) [86], which is widespread in eukaryotes and serves as a quality control mechanism [132]. NMD is coupled with AS to regulate the levels of functional mRNA transcripts through the specific degradation of alternatively spliced isoforms possessing a premature stop codon [133]. However, it remains unclear that how the potential NMD targets derived from AS of R genes escape being destroyed. Although its functional roles are not clear and few targets have been identified, NMD regulation in disease resistance has been documented. nmd mutants display stunted phenotypes and curled leaves that resemble those of mutants with constitutive activation of defense responses [134], and the majority of transcripts enriched in nmd mutants are associated with the pathogen response [134].
The AS transcripts of R genes such as N and RPS4 are of low abundance in the absence of their corresponding effectors. It was first assumed that NMD activity was repressed during pathogen infection, resulting in the accumulation of alternative R gene transcripts [135], which is consistent with the weak NMD activity observed by RNA-seq analysis in P. syringae-infected Arabidopsis [15]. Alternatively, the R gene transcript isoforms possessing premature stop codons may be insensitive to NMD. Generally, mRNAs targeted for NMD have uORFs or a larger 3'-UTR region [136]. However, many transcripts displaying intron retention are not sensitive to NMD, although they appear to have the characteristics of transcripts affected by NMD. Whatever the mechanism, more evidence is required to clarify the molecular machinery that suppresses NMD of alternatively spliced transcripts of R genes.
The cloning of the MOS genes shed light onto spliceosomal regulation in the AS of R genes. However, the splicing factors and RNA-binding proteins responsible for pre-mRNA splicing of R genes other than SNC1 and RPS4 are completely unknown. One challenge is that mutations in these factors may cause inconspicuous phenotypes, because AS could be unnecessary for full disease resistance. In addition, even if defense signaling requires AS of an R gene, the mutant would likely grow normally in the absence of the pathogen expressing its cognate effector. We therefore have a long way to go before the full picture of regulation of R gene AS is revealed. Continuous advances in genomics, bioinformatics, transcriptomics, phenomics, and sequencing technologies, should facilitate our explorations of the complexity for generation and contribution of AS of R genes and elucidate new avenues to modify plant innate immunity.
Agriculture of University of Kentucky. This article is published with the approval of the Director of the Kentucky Agricultural Experiment Station.

Author Contributions
S.Y. and H.Z. wrote the manuscript; F.T and S.Y performed the experiments to characterize the role of alternative splicing in RCT1-mediated disease resistance in Medicago truncatula.