Plant resistance (R) proteins protect cells from infections through recognizing effector molecules produced by pathogens and initiating downstream defense cascades. To mount proper immune responses, the expression of R genes has to be tightly controlled transcriptionally and post-transcriptionally. Intriguingly, alternative splicing of the R genes of the nucleotide binding leucine-rich repeat (NB-LRR) type was observed in different plant species, but its regulatory mechanism remains elusive. Here, we report the positional cloning and functional analysis of modifier of snc1,12 (mos12-1), a partial loss-of-function mutant that can suppress the constitutive defense responses conferred by the gain-of-function R gene mutant suppressor of npr1-1constitutive 1 (snc1). MOS12 encodes an arginine-rich protein that is homologous to human cyclin L. A null allele of mos12-2 is lethal, suggesting it has a vital role in plant growth and development. MOS12 localizes to the nucleus, and the mos12-1 mutation results in altered splicing patterns of SNC1 and RPS4, indicating that MOS12 is required for the proper splicing of target R genes. MOS12 co-immunoprecipitates with MOS4, indicating that MOS12 associates with the MOS4-associated complex (MAC). Accordingly, splicing patterns of SNC1 and RPS4 are changed in most MAC core mutants. Our study highlights the contribution of MOS12 and the MAC in the alternative splicing of R genes, providing regulatory details on how alternative splicing is used to fine-tune R gene expression in plant immunity.
Plants have evolved sophisticated immune systems to intercept the invading pathogens that threaten normal growth and development. The first layer of immunity is usually mediated through receptor-like kinases (RLKs) at the plasma membrane, which recognize conserved pathogen-associated molecular patterns (PAMPs) such as flagellin or EF-Tu of bacteria and chitin of fungi, leading to PAMP-triggered immunity (PTI) that can halt further pathogen proliferation (Jones and Dangl, 2006). The second layer of immunity is mediated by resistance (R) proteins, which can specifically recognize pathogen-secreted effectors that often dampen PTI and establish effector-triggered immunity (ETI) (Staskawicz et al., 1995; Jones and Dangl, 2006). ETI often culminates in a localized hypersensitive response (HR) that efficiently restricts pathogen growth (Hammond-Kosack and Jones, 1996; Chisholm et al., 2006). Most R proteins have a predicted NB-LRR structure, which contains leucine-rich repeats (LRRs) at the C terminus, a central domain with a nucleotide-binding site (NB) and either a Toll/Interleukin-1-receptor-like (TIR) or a coiled-coil (CC) domain at the N terminus (Belkhadir et al., 2004). TIR-type R protein-mediated immunity usually depends on ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1), PHYTOALEXIN DEFICIENT 4 (PAD4) and SENESCENCE-ASSOCIATED GENE 101 (SAG101), which all share homology with eukaryotic lipases, whereas the CC-type R proteins typically rely on NON-RACE-SPECIFIC DISEASE RESISTANCE 1 (NDR1) (Century et al., 1995; Aarts et al., 1998; Falk et al., 1999; Jirage et al., 1999; Feys et al., 2001, 2005). Strikingly, these plant R proteins share structural similarities with animal NB domain and LRR-containing (NLR) innate immunity sensors, such as Nod-like proteins (Jones and Dangl, 2006).
In Arabidopsis, the gain-of-function mutant suppressor of npr1-1constitutive 1 (snc1) constitutively expresses PATHOGENESIS RELATED (PR) defense marker genes, and exhibits enhanced disease resistance against virulent Pseudomonas syringae pv. maculicola (Psm) ES4326 bacteria and the oomycete pathogen Hyaloperonospora arabidopsidis (Ha) Noco2 (Li et al., 2001). SNC1 encodes a TIR-type NLR R protein with a point mutation in the linker region between the NB and LRR, resulting in a Glu→Lys change and rendering the R protein constitutively active (Zhang et al., 2003). To identify components required for TIR-type R protein-mediated immunity, forward-genetic screens were employed to identify mutants that can suppress the autoimmune phenotypes of snc1. Multiple mutagens including fast neutron, T-DNA and ethyl methane sulphonate (EMS) were used in independent modifier of snc1 (MOS) screens, and revealed the importance of several cellular and molecular events required for snc1 signaling, including nucleocytoplasmic trafficking, RNA processing and protein modification (Monaghan et al., 2010). For example, mos4-1 completely suppresses all autoimmune phenotypes of snc1. MOS4, a small protein with coiled-coil domain, is part of an evolutionary conserved spliceosome-associated MOS4-associated complex (MAC). The yeast counterpart of the MAC is the NineTeen Complex (NTC), and the human one is called CDC5L or PRP19 complex. The MAC also contains other core components, AtCDC5, an atypical R2R3 Myb transcription factor, Pleiotropic Regulatory Locus 1 (PRL1), a WD-40 protein, and MAC3A and MAC3B, two functionally redundant U-box E3 ubiquitin ligases that are homologous to PRP19 (Palma et al., 2007; Monaghan et al., 2009). Although the MAC is closely associated with the spliceosome, and mutants of all MAC core components exhibit enhanced disease susceptibility. The detailed function of the MAC in plant immunity is unclear.
A number of R genes, including N of tobacco (Dinesh-Kumar and Baker, 2000), MLAs of barley (Halterman et al., 2003), RPS4 (Gassmann et al., 1999; Zhang and Gassmann, 2003, 2007), RPP5, SNC1, and RPS6 (Parker et al., 1997; Yi and Richards, 2007; Kim et al., 2009) of Arabidopsis were reported to be alternatively spliced, although the regulatory mechanisms for R gene alternative splicing remains elusive. Here, we report the identification and characterization of MOS12, which encodes a protein with high homology to cyclin L of mouse and human. Like cyclin L, MOS12 contains two cyclin domains at the N terminus and a C-terminal serine–arginine (SR) domain that is typical for splicing factors belonging to the SR protein family. In mos12-1, the splicing patterns of SNC1 and RPS4 are altered, and the splicing defect of SNC1 results in reduced levels of the R protein. In addition, we found that MOS12 associates with the MAC in planta, and that components of the MAC, including AtCDC5, MOS4 and MAC3, also contribute to the proper splicing of SNC1 and RPS4. This study provides the emerging regulatory details of alternative splicing of certain R genes, which seems to be a critical step in controlling defense outputs.
Identification of the mos12-1 mutant
Mutant snc1 was originally identified as a genetic suppressor of npr1-1 and constitutively activates plant defense responses without pathogen infection (Li et al., 2001). snc1 plants exhibit a stunted stature, have curled leaves, constitutively express PR genes and display enhanced resistance to virulent pathogens. To dissect the components required for TIR-type R-protein signaling, forward-genetic screens were designed to search for mutants that can suppress snc1-mediated autoimmunity. mos12-1 was obtained from an EMS mutagenized snc1 npr1 population (Xu et al., 2011). snc1 npr1 carries the pPR2-GUS reporter gene, where PR2 expression can be monitored by GUS staining (Li et al., 2001). As shown in Figure 1(a), the mos12-1 snc1 npr1 plant is much larger than the snc1 npr1 double mutant, and suppresses the stunted growth and curly leaf morphology of snc1 npr1. When mos12-1 snc1 npr1 was backcrossed with snc1, F1 plants were snc1-like, indicating that mos12-1 is recessive. The constitutive expression of PR-1 and PR-2 was greatly reduced by mos12-1 in snc1 npr1, as analysed by GUS staining (Figure 1b) and RT-PCR (Figure 1c). To determine whether mos12-1 also altered resistance against virulent pathogens, mos12-1 snc1 npr1 together with snc1 npr1 and wild-type plants were challenged with the oomycete Ha Noco2 or the bacterial pathogen Psm ES4326. As shown in Figure 1(d,e), the enhanced resistance responses of snc1 npr1 were abolished in the triple mutant. The bacterial growth in mos12-1snc1 npr1 was even higher than that in the wild type (Figure 1e). Taken together, mos12-1 suppresses all aspects of snc1-mediated autoimmunity.
Map-based cloning of mos12-1
To identify mos12-1, a positional cloning approach was applied. Mutant mos12-1 snc1 npr1 (in the Columbia, Col, background) was crossed with Landsberg erecta (Ler) that had been previously introgressed with snc1, Ler-snc1 (Zhang and Li, 2005). Using 24 F2 plants that suppressed snc1 morphology, the mos12-1 mutation was mapped to the middle of chromosome 2 by linkage analysis. The mutation was further flanked between markers T22F11 and T18O18 using 94 mos12-1-like F2 plants. An additional 82 recombinants were collected between markers T22F11 and T18O18 using another 887 F2 progeny. These recombinants were further analysed using markers between T22F11 and T18O18 (Table S1). The mutation was eventually narrowed down between markers S9T8 and C2F4, with a distance of 67 kb. This region was located on BAC clone J9J22 (Figure 2a). Overlapping fragments covering the coding sequences of the region were PCR amplified from mos12-1 snc1 npr1 mutant genomic DNA and sequenced. One G→A transition in At2g26430 was detected when the mutant sequence was compared with the Arabidopsis reference genome sequence (Figure 2b). This point mutation occurred at the exon–intron boundary, which is part of the splice site consensus sequence (Figure 2c).
To test whether the G→A mutation affected the mRNA of At2g26430, the cDNA of At2g26430 was amplified by RT-PCR using RNA extracted from mos12-1 snc1 npr1 plants. Comparison of the cDNA sequence between mos12-1 and the wild type showed that the splicing pattern of At2g26430 was altered in mos12-1, which included five extra nucleotides (Figure 2c). As a consequence, the reading frame is shifted, truncating the protein with an earlier stop codon (Figure 2d). To confirm that At2g26430 is MOS12, we transformed mos12-1 snc1 npr1 plants with a genomic clone of At2g26430 tagged with a FLAG epitope at its C terminus. Among 12 T1 transgenic plants, eight exhibited snc1 npr1 morphology (Figure S1a), suggesting that At2g26430 complemented mos12-1 and MOS12 is indeed At2g26430.
We next generated the mos12-1 single mutant by crossing mos12-1 snc1 npr1 with the wild type, followed by allele-specific genotyping in F2. The mos12-1 single mutant is slightly smaller than the wild type with smaller and lighter green leaves (Figure S1b). A second T-DNA insertion allele of mos12, SALK_013352, which was named mos12-2, was obtained from the Arabidopsis Biological Resource Center (ABRC). mos12-2 carries a T-DNA insertion in the fourth exon of At2g26430 (Figure 2b). Among the 36 plants grown, none were homozygous for mos12-2, whereas heterozygous plants resembled the wild type. From 32 progeny of two independent plants heterozygous for the T-DNA insertion, no homozygous plants were obtained, whereas 24 plants were heterozygous. These data suggest that homozygous plants of mos12-2 are lethal (expected ratio 2 : 1; χ2 = 1.00; 0.1 < P < 0.5). To test whether the lethality of mos12-2 is allelic to mos12-1, a mos12-2 heterozygous plant was crossed with homozygous mos12-1. In the F1 progeny, two types of plants segregated at about a 1 : 1 ratio: one wild type-like whereas the other resembled mos12-1, but was smaller in size (Figure S1b). Genotyping of these F1 plants revealed that all wild type-like plants did not carry the T-DNA insertion, whereas all smaller plants were heterozygous for mos12-1 and mos12-2, indicating that the lethality of mos12-2 is allelic to mos12-1. As mos12-2 is lethal but mos12-1 is not, we concluded that mos12-1 is a partial loss-of-function allele of MOS12.
MOS12 has three different splice variants according to the expressed sequence tag (EST) information from the Arabidopsis Information Resource (TAIR) (Figure 2b). To investigate whether the splicing pattern of MOS12 is changed in mos12-1, primers flanking the alternatively spliced region were used to amplify the transcript variants in the wild type and mos12-1 (Figure S2a). The splicing pattern of MOS12 is slightly affected by mos12-1, with the long transcript variant being more abundant and the short variants slightly less in mos12-1, compared with the wild type (Figure S2b). We also tested whether the splicing pattern of MOS12 is changed by pathogen infection. As shown in Figure S2(c,d), the splicing pattern of MOS12 in mos12-1 is not significantly altered with induction from the virulent pathogen Psm ES4326 or the avirulent pathogen Pseudomonas syringar pv. tomato (Pst) avrRPS4.
Basic Local Alignment Search Tool (BLAST; http://www.ncbi.nlm.gov/Blast.cgi) analysis revealed that MOS12 encodes a protein with two conserved cyclin domains at the N terminus. It is rich in arginine, alternating with serine, glutamate and/or aspartate residues at the C terminus (Figure 2d). Phylogenetic analysis and protein alignment of MOS12 and its homologs revealed that MOS12 is a conserved protein with homologs in eukaryotic species (Figures S3 and S4). The closest MOS12 homolog in humans is cyclin L.
MOS12 localizes to the nucleus
To better understand the function of MOS12, we first investigated its subcellular localization. The sequence of GFP was fused in-frame with genomic MOS12 and expressed as a C-terminal tag under the control of its endogenous promoter in the mos12-1 snc1 npr1 triple mutant. The transgene was able to complement the morphological phenotype of mos12-1 snc1 npr1 (Figure 3a), and restored the constitutive defense against virulent Ha Noco2 (Figure 3b), suggesting that the MOS12-GFP fusion protein localizes to the correct subcellular compartment, and is as functional as MOS12. MOS12-GFP green fluorescence was observed in the nucleus when leaf epidermal cells of transgenic plants were examined by confocal fluorescence microscopy (Figure 3c), indicating that MOS12 is a nuclear protein. The localization of MOS12-GFP agrees with that of the human homolog cyclin L, which also localizes to the nucleus (Loyer et al., 2008).
MOS12 is required for proper splicing of SNC1
As cyclin L is involved in pre-mRNA splicing, we investigated the involvement of MOS12 in regulating splicing and the transcript level of SNC1 using the mos12-1 single mutant. SNC1 was suspected to be a MOS12 target as snc1-mediated autoimmunity is suppressed by mos12-1 and SNC1 is alternatively spliced (Yi and Richards, 2007; Xu et al., 2011). To determine whether mos12-1 affects SNC1 transcript, the transcript level of SNC1 was analyzed by semi-quantitative RT-PCR. The snc1 transcript level was decreased in the mos12 snc1 npr1 triple mutant when compared with the snc1 npr1 double mutant (Figure 4a). Consistent with this, the SNC1 level in mos12-1 was also slightly reduced compared with the wild type (Figure 4a).
SNC1 is alternatively spliced, with intron 2 and 3 retained or removed (Yi and Richards, 2007; Xu et al., 2011). To test whether the SNC1 splicing pattern is altered in mos12-1, multiple primer combinations were used to amplify the SNC1 transcript variants in the mos12-1 single mutant and in the mos12-1 snc1 npr1 triple mutant (Figure 4b,c). Three extra bands representing alternatively spliced variants of SNC1 were amplified much more strongly using primers F203 and R206 in the mos12-1 single mutant and the mos12-1 snc1 npr1 triple mutant, compared with the wild type and the snc1 npr1 double mutant (Figure 4c), suggesting that either intron 2 or 3, or both, are retained more in the mos12-1 background. Using primers F203 and R204, the retention of intron 2 could be specifically observed in the mos12-1 single mutant and in the mos12-1 snc1 npr1 triple mutant (Figure 4c). The retention of intron 3 was also specifically detected in the mos12-1 mutant with primers F205 and R206 (Figure 4c). To determine whether the protein level of SNC1 was consequently affected, SNC1 protein abundance was examined by western blot analysis of the mos12-1 single mutant and the mos12 snc1 npr1 triple mutant. As shown in Figure 4(d), the SNC1 protein level was decreased in the mos12-1 single mutant and in the mos12-1 snc1 npr1 triple mutant, as compared with the wild type and the snc1 npr1 double mutant, respectively. Taken together, we conclude that the mutation in mos12-1 affects the SNC1 splicing pattern and SNC1 transcript level, which further affects SNC1 protein production.
MOS12 is required for both basal and RPS4-mediated disease resistance, but not for either RPS6- or RPS2-mediated immunity
To determine whether the basal defense response was altered in the mos12-1 plants, mos12-1 single mutant plants were challenged with the virulent pathogen Psm ES4326 at a subclinical concentration of OD600 = 0.0001. At this low dose of inoculum, wild-type Col plants do not support strong growth of the bacteria, whereas mutants with enhanced disease susceptibility do. For instance, mutant Col eds1-2, in which the function of a key component of plant immunity EDS1 is abolished, exhibits a 100–1000-fold greater proliferation of bacteria than the wild type (Figure 5a). After 3 days, about 10-fold more bacterial growth was observed in the mos12-1 single mutant than in the wild type (Figure 5a). Therefore, MOS12 contributes to basal defense.
As another TIR-type R gene RPS4 was reported to be alternatively spliced (Gassmann et al., 1999; Zhang and Gassmann, 2003, 2007; Xu et al., 2011), we investigated the contribution of MOS12 to RPS4-mediated immunity. When mos12-1 plants were inoculated with avirulent Pst DC3000 strains carrying the effector avrRPS4 that is recognized by the TIR-type R protein RPS4 (Hinsch and Staskawicz, 1996; Gassmann et al., 1999), a roughly fivefold higher growth of bacteria was detected, compared with the wild type (Figure 5b). mos12-1 did not exhibit as severe susceptibility as Col eds1-2, which specifically abolishes TIR-type R protein-mediated defense responses. Therefore, MOS12 also seems to contribute to disease resistance conferred by RPS4.
We then tested whether the splicing pattern of RPS4 was altered in mos12-1. RPS4 is alternatively spliced, with intron 2 and 3 retained or removed (Zhang and Gassmann, 2003, 2007; Xu et al., 2011). Primers flanking the introns were used to amplify the transcript variants (Figure 5c). As shown in Figure 5(d), the splicing pattern of RPS4 is significantly changed in mos12-1 compared with the wild type. Thus MOS12 is required for proper slicing of not only SNC1, but also RPS4.
Besides RPS4, another TIR-type R gene RPS6 was also reported to be alternatively spliced (Kim et al., 2009). When we challenged mos12-1 with avirulent Pst DC3000 strains carrying the effector avrHopA1 that is recognized by the TIR-type R protein RPS6 (Gassmann, 2005), no obvious difference of bacterial growth was observed between the wild type and mos12-1 (Figure S5a), which indicates that RPS6-mediated defense response is not compromised in the mos12-1 mutant. In contrast, high bacterial growth was observed on Col eds1-2 (Figure S5a), which specifically abolishes TIR-type R protein mediated defense responses. When we examined the alternative splicing pattern of RPS6 in mos12-1, no significant difference was detected (Figure S5b), indicating that MOS12 is not required for the proper splicing of RPS6.
Most known CC-type R genes in Arabidopsis do not carry introns. To test whether CC-type R protein-mediated defense is affected in mos12-1, we inoculated mos12-1 with avirulent Pst DC3000 strains carrying the effector avrRPT2 that is recognized by the CC-type R protein RPS2 (Yu et al., 1993). The bacteria grew similarly in mos12-1 as in the wild type, whereas much more growth was observed in ndr1-1, where many CC-type R protein-mediated defense responses are abolished (Century et al., 1995; Aarts et al., 1998; Falk et al., 1999; Jirage et al., 1999; Feys et al., 2001, 2005). The infection data suggest that MOS12 is not required for the RPS2-mediated defense response.
Taken together, it seems that only the splicing of specific R genes is affected in mos12-1, most of which are probably of the TIR type.
Double mutant analysis between mos12-1 and mos4 mutants
MOS12 is a single-copy gene in Arabidopsis. As a homolog of cyclin L that is involved in splicing, it is not surprising that the mos12-2 null allele is lethal. However, mos12-1 plants can grow and reproduce normally. This unique partial loss-of-function mutant provides a good opportunity to genetically test whether a protein candidate might be involved in mRNA splicing. We reasoned that if the double mutant of the tested candidate gene and mos12-1 exhibit lethality, it may be a good indication that this candidate is involved in splicing, as its mutation is probably abolishing the remaining splicing activity in mos12-1.
The known mRNA export mutant mos3-2 looks similar to wild-type plants except that it flowers early. The flowering time of mos12-1 is the same as the wild type, but it is distinguished by smaller and lighter green leaves. When mos12-1 was crossed with mos3-2, the double mutant plants are viable and exhibit an additive phenotype resembling both mos12-1 and mos3-2 (Figure S6), with smaller and lighter leaves and early flowering, suggesting that MOS3 is not involved in splicing, and that the combination of mos12 with a mutant carrying a mutation in RNA processing steps other than splicing would not yield a lethality phenotype.
MOS4 is one of the core components of the MAC that associates with the spliceosome. However, qualitative splicing defects in several alternatively spliced genes were not detected in the MAC core mutants mos4-1, Atcdc5, prl1 and mac3a mac3b (Palma et al., 2007; Monaghan et al., 2009). Although we did notice subtle quantitative differences in the alternative splicing patterns of certain genes, such as RPS4 in the mos4 mutant, we were not confident in drawing the conclusion that the MAC contributes to the alternative splicing of the R genes. When we crossed mos12-1 with mos4-1, no double mutant could be found in 72 F2 plants. Among 80 F3 progeny from two independent F2 plants that were heterozygous for mos12-1 and homozygous for mos4-1, no double mutant could be identified. These data suggest that the mos12-1 mos4-1 double mutant is lethal, and that MOS4 is probably also involved in splicing. Mutation in MOS4 probably abolishes the remaining splicing activities in mos12-1.
As the MAC interacts with the spliceosome and the SR domain of MOS12 is predicted to also associate with the spliceosome, we tested the interaction between the MAC and MOS12 through co-immunoprecipitation analysis. When MOS4-HA transgenic plants in the mos4-1 mutant background (Palma et al., 2007; Monaghan et al., 2009) were subject to immunoprecipitation using anti-HA microbeads, not only the MAC core component AtCDC5, but also MOS12, was detected in the elution fractions (Figure 6a). These data suggest that MOS12 associates with the MAC in planta, possibly through the spliceosome as a common interaction partner.
As the SNC1 and RPS4 splicing patterns are altered in the mos12-1 mutant, we carefully analysed further whether they are also changed in MAC core component mutants, including mos4-1, Atcdc5, mac3a mac3b and prl1-1, using the same primer combinations used for the analysis of the mos12-1 mutant in Figures 4 and 5. As shown in Figure 6(b), the splicing pattern of SNC1 is changed significantly in the MAC mutants mos4-1, Atcdc5 and mac3a mac3b, whereas that of prl1-1 is similar to the wild type. Similarly, the splicing pattern of RPS4 is considerably changed in mos4-1, Atcdc5 and mac3a mac3b, whereas the RPS4 splicing pattern in prl1-1 is similar to the wild type (Figure 6c). These data show that, like MOS12, the MAC core components MOS4, AtCDC5 and MAC3 also regulate the proper splicing of target R genes, including SNC1 and RPS4.
In this study we have identified and characterized the mutant mos12-1, which suppresses the autoimmune phenotypes of snc1 npr1 through the alteration of the SNC1 splicing pattern. MOS12 encodes a protein similar to human cyclin L that plays a critical role in splicing. It contains two conserved cyclin domains at the N terminus and an SR-rich domain at the C terminus. Our functional analysis further revealed that MOS12 associates with the spliceosome-associated MAC in the nucleus, both of which are required for the proper splicing of R genes, such as RPS4 and SNC1. It provides regulatory details on R-gene splicing, which seems to be a critical step in controlling the defense output of certain R genes.
In eukaryotes many genes are discontinuous, with exons interrupted by non-coding introns. Splicing, which excises introns and ligates exons, is an indispensable step in gene expression. This process is orchestrated by a highly conserved ribonucleoprotein (RNP) spliceosome complex assembled from small nuclear RNP particles (U-snRNPs) and numerous protein factors (Will and Lührmann, 2001). Different classes of proteins, such as RNA-dependent ATPases, RNA helicases, protein kinases and RNA binding proteins are involved in the regulation of splicing. Splicing has primarily been studied in yeast and human using biochemical approaches. Most splicing factors have not yet been studied experimentally in plants. Nevertheless, homologs of most spliceosome constituents and splicing-related proteins have been identified in Arabidopsis using computational analysis, suggesting an evolutionarily conserved nature of splicing and possibly its regulation (Wang and Brendel, 2004).
MOS12 shares high homology with human cyclin L. The biochemical functions of cyclins have been extensively explored since they were first found to play important roles in cell cycle regulation. Two major groups of cyclins have distinct functions. Cyclins A, B, D and E work with their cyclin-dependent kinase partners, and play a pivotal role in the regulation of the cell cycle. Another group, including cyclins C, H, K, L and T, tend to play roles in transcriptional regulation (Bregman et al., 2000). Cyclin L, including cyclins L1 (Dickinson et al., 2002) and L2 (de Graaf et al., 2004), contains common cyclin domains, as do other members of the cyclin family. They also have a unique SR domain with a high content of the dipeptides SR or RS, which are frequently found in metazoan SR proteins involved in pre-mRNA splicing (Shepard and Hertel, 2009). Cyclin L appears to associate with RNA Pol II and CDKs through its cyclin domain, and with splicing factors such as SC-35 through its SR domains (Berke et al., 2001; Dickinson et al., 2002; de Graaf et al., 2004; Yang et al., 2004; Chen et al., 2006). The interaction between cyclin L and splicing factors and the fact that cyclin L can stimulate splicing in vitro suggests that it is involved in mRNA splicing (Dickinson et al., 2002; Loyer et al., 2008). MOS12 contains a rather atypical SR domain where arginines alternate with serine, glutamate and/or aspartate residues. Nevertheless, the domain rich in these dipeptides is probably still responsible for mediating related protein–protein or protein–RNA interactions, based on their similar amino acid properties as SR or RS repeats. This is further supported by the fact that both typical and atypical SR domain-containing proteins can be recognized by the antibody mAb 16H3 (Neugebauer et al., 1995). The homology between MOS12 and cyclin L indicates that MOS12 may also function in splicing through an association with one of the CDKs and splicing factors via its interaction interface in the cyclin domains.
From a previous cDNA library screen to search for Arabidopsis components that would confer salt tolerance to yeast cells, Forment et al. (2002) found that the overexpression of MOS12 (previously named RCY1) is able to grant salt tolerance to Saccharomyces. This tolerance appears to rely on the phosphorylation of MOS12. It is proposed that MOS12 might be able to stimulate splicing and/or other steps of mRNA metabolism compromised by salt stress, probably interacting with protein components of the pre-mRNA processing machinery. However, the exact mechanism of MOS12 in salt tolerance remains unclear.
For many genes in multicellular organisms, the same pre-mRNA can produce different mature transcripts through the selective removal of different introns and joining exons. This alternative splicing can potentially lead to structurally and functionally distinct proteins. By analyzing mRNA-sequence and EST data, it was estimated that over 95% of multi-exon genes in human and 42% intron-containing genes in Arabidopsis may undergo alternative splicing (Pan et al., 2008; Filichkin et al., 2010). The ubiquity of alternative splicing has extensively enriched the diversity of the proteome in multicellular organisms. In plants, alternative splicing is often associated with specific types of tissue, developmental stages or environmental conditions such as stresses. One intriguing phenomena is the alternative splicing of R genes (Gassmann, 2008). N in tobacco was the first reported R gene to be alternatively spliced. Splice variants of N can be induced by pathogen attack, and seem to be crucial for resistance (Dinesh-Kumar and Baker, 2000). The alternative splicing of RPS4, SNC1, RPP5 and RPS6 (Parker et al., 1997; Zhang and Gassmann, 2003; Yi and Richards, 2007; Kim et al., 2009) has also been detected in Arabidopsis. In particular, RPS4 undergoes dynamic splice-form changes during the resistance response (Zhang and Gassmann, 2007). Some other NB-LRR genes, such as MLAs in Hordeum vulgare (barley; Halterman et al., 2003), Bs4 in Solanum lycopersicum (tomato; Schornack et al., 2004), and L6 and M in Linum usitatissimum (flax; Anderson et al., 1997; Ayliffe et al., 1999; Schmidt et al., 2007) can undergo alternative splicing as well. These studies indicate that the alternative splicing of R genes may play a pivotal role in the regulation of plant immunity, even though little is known about its regulatory mechanism. Here, we found that MOS12 contributes to immunity at the level of alternative splicing regulation. Splicing patterns of both SNC1 and RPS4 are changed in mos12-1. The altered splicing patterns of SNC1 in mos12-1 probably contribute to its ability to suppress snc1 autoimmunity. In addition, basal defense against Psm ES4326 is compromised in the mos12-1 single mutant, although it is unclear how MOS12 affects basal defense. One possibility is that mos12-1 may disturb the regular splicing pattern of key defense regulators required for basal resistance. Alternatively, mos12-1 may cause splicing defects of multiple R genes that culminate in a basal resistance defect against Psm ES4326.
One big puzzle is the specificity of MOS12. As a null allele of mos12, such as mos12-2, is lethal, it could be an essential component of general splicing, and is required for the splicing of all genes with introns. However, our analysis on mos12-1 seems to indicate that only specific genes in mos12-1 are altered in terms of their splicing pattern. mos12-1 decreases the intron-retained transcript variant of AtSR1/SRp34, one known SR gene in Arabidopsis, but does not seem to affect the splicing patterns of another SR gene, AtSRp30, nor the U1snRNP 70K (U1-70K) gene, which is part of the spliceosome (Figure S7a). In addition, mos12-1 specifically reduced the intron-containing transcript variants of JAZ2, but not JAZ10 (Figure S7a), both of which are essential components of JA signaling (Fonseca et al., 2009). These data suggest that mos12-1 exhibits distinguished specificity on alternative splicing of certain genes, as it does on R-gene splicing, in which it significantly affects the alternative splicing of SNC1 and RPS4, but not of RPS6. Besides alternative splicing, we could not rule out the possibility that mos12-1 also affects constitutive splicing. However, the transcript level of PAD4, which is not alternatively spliced, is not changed in mos12-1, suggesting that the constitutive splicing of PAD4 was not altered by mos12-1 (Figure S7b). Future analysis of the full-genome splicing patterns in mos12-1, using reliable methods like tiling arrays or transcriptome sequencing, may shed light on which defense-gene splicing patterns are altered in this unique mutant.
In contrast to the partial loss-of-function allele mos12-1, the null mos12-2 allele exhibits lethality. This is not surprising considering its predicted essential role in splicing as a homolog of cyclin L. Mutations in splice factors often result in catastrophic phenotypes or death. For example, mutations in the splice factor U2AF in Drosophila and Sfrs10 in mice cause lethality (Rudner et al., 1996; Mende et al., 2010). Mutations in the Arabidopsis splice factors AtCAF or LACHESIS (LIS) also result in embryo lethality (Asakura and Barkan, 2006; Gross-Hardt et al., 2007). In mos12-1, the G→A mutation causes a change at a splice junction site, leading to a frame shift, and produces a premature termination codon (PTC). However, the detection of MOS12 transcript in mos12-1 indicates that it successfully evades the nonsense-mediated decay (NMD), which is an mRNA surveillance mechanism that typically degrades transcripts that contain PTCs that code C terminus-truncated proteins that can be toxic to the cells through a dominant-negative or gain-of-function effect. NMD is essential for eliminating PTC containing mRNA generated by errors in transcription and pre-mRNA splicing. It is reported that only PTC localizing more than 55 bp upstream of the last exon–exon junction can be distinguished by NMD in a mammalian NMD surveillance system (Khajavi et al., 2006). Therefore, it is not surprising that the PTC-containing transcripts in mos12-1 successfully escape NMD because the PTC localizes in the last exon. However, the truncated protein of MOS12-1 seems partially functional, as mos12-1 plants exhibit relatively normal growth and development. A possible explanation for the partial loss-of-function phenotype of mos12-1 is that the truncated protein in mos12-1 still carries the most conserved cyclin domains. The protein may remain partially functional during splicing, giving rise to an almost normal plant that is slightly stunted. It is intriguing that the mutation in mos12-1 causes differences in the splicing pattern of MOS12 (Figure S2). One explanation is that the splicing of MOS12 is under the control of itself. When the MOS12 protein is truncated in mos12-1, the splicing pattern of MOS12 is altered. Another possibility is that mutation in mos12-1 leads to altered functions of components regulating the splicing of MOS12. Thus the splicing pattern of MOS12 is changed indirectly. The third possibility could be that the point mutation in mos12-1 is recognized incorrectly by the splicing machinery, and ultimately changes its splicing pattern.
Our study on MOS12 revealed its function in splicing, which is one major step of RNA processing that includes transcription, 5′ capping, 3′ poly-adenylation, splicing, mRNA export and mRNA quality control. Each stage during RNA processing is tightly regulated and contributes to the number and type of final functional transcripts of a gene. In Arabidopsis, mutants in different components of the RNA processing machinery seem to have their own phenotypic signatures, and mutants in the same RNA processing step tend to have similar phenotypes. For instance, mutants of MOS3, a putative homolog of nucleoporin 96, and LOS4, an RNA helicase, resemble each other and flower earlier than the wild type. Both MOS3 and LOS4 are involved in mRNA export (Gong et al., 2005; Zhang and Li, 2005). Double-mutant analysis between an unknown component of RNA processing with a known component may help us to define its function. For example, the mos12-1 mos3-2 double mutant is viable and exhibits an additive phenotype, resembling both mos12-1 and mos3-2, suggesting that the steps in mRNA processing (i.e. splicing and export) that these two encoded proteins are involved in are probably not the same. Using this logic, the mos12-1 mos4-1 double mutant is lethal, a phenotype shared with the mos12-2 null allele or any of the MAC double mutant combinations (Palma et al., 2007; Monaghan et al., 2009), and we were able to deduce that MOS4 probably contributes to splicing by itself. Altered SNC1 and RPS4 splicing patterns in the MAC core component mutants mos4, Atcdc5 and mac3a mac3b further supported this hypothesis (Figure 6). Interestingly, the SNC1 splicing pattern is not changed in prl1, which could be explained either by a masking effect provided by the partially redundant PRL2 that is highly homologous to PRL1, or by the multifunctionality of PRL1, other than acting as a scaffold in the MAC. Our double mutant analysis helped to reveal the contribution of single MAC components to splicing, which in the past has not been possible without the mos12-1 mutant. For future studies, the mos12-1 mutant will provide a unique tool for plant biologists to genetically test whether a candidate protein is involved in splicing.
It has been over 10 years since the first discovery of alternative R-gene splicing in plants. Our studies of MOS12 reinforce the important regulatory function of alternative splicing for plant immunity. MOS12 and the MAC are found to contribute to the proper splicing of target R genes. Future in-depth mechanistic analysis will help reveal how these regulatory components respond to pathogen attacks and how R-gene alternative splicing contributes to R-protein expression and function.
Plant growth conditions and mutant screen
All plants were grown at 22°C under a 16-h light/8-h dark or 10-h light/14-h dark regime. The MOS screen using EMS was carried out in a similar fashion as the previously described screen, using fast neutrons, except that EMS was used as a mutagen (Zhang et al., 2005). Briefly, M2 plants were planted and screened for the suppression of the dwarf phenotype of snc1 npr1. Plants that were no longer dwarfed were further analysed by GUS staining to test whether constitutive pPR2::GUS defense gene expression was also suppressed.
Gene expression analysis
RNA samples were extracted from about 0.1 g of 2-week-old MS plate-grown seedlings using the Totally RNA kit (Ambion Cat# AM1910, now part of Invitrogen, http://www.invitrogen.com/site/us/en/home/brands/ambion.html). A 0.4-μg portion of RNA was reverse transcribed to generate cDNA using SuperScript II reverse transcriptase (Invitrogen, http://www.invitrogen.com). cDNA samples were initially normalized with ACTIN by real-time PCR using the QuantiFAST SYBR Green PCR kit (Qiagen, http://www.qiagen.com). The cDNA were subsequently amplified by PCR using 94°C for 2 min and cycles of 94°C for 15 sec, 58°C for 30 sec and 68°C for 1 min. The extension time was adjusted to 1 min and 30 sec with primers F203 and R206. The primers used for amplification of PR-1, PR-2 and ACTIN were described previously (Cheng et al., 2009). The primer sequences for SNC1 transcript analysis are as follows: snc1-F, 5′-GTGGAGTTCCCATCTGAACATC-3′; snc1-R, 5′-CGTTCAAAGGCATGCGTAATCTG-3′; F203, 5′-AGGGAAGGACCAAAGAATGG-3′; R204, 5′-TGAGGTAGATCCCCGTAATA-3′; F205, 5′-AAATTG GTTATTACGGGGATC-3′; and R206, 5′-AATTGTTGGAATACCTCAAATT-3′. The primer sequences for RPS4 transcript variants analysis are as follows: RPS4-F, 5′-CTGTGGCTCCATCAACACAT-3′; RPS4-R, 5′-GATCGACCCACCTTAAGCAT-3′.
The positional cloning of mos12-1 was performed as previously described (Zhang et al., 2005). The markers used to map mos12-1 were designed according to the insertion, deletion or single nucleotide polymorphisms between the genomic sequences of Col and Ler ecotypes, provided by Monsanto on TAIR (http://www.arabidopsis.org; Table S1).
Construction of plasmid
The genomic sequence covering the MOS12 coding region without the stop codon plus 1.9 kb 5′ upstream sequence was amplified by PCR with BamHI and SalI restriction enzyme sites, introduced separately at the two primer ends. The fragment was then ligated to the modified pCAMBIA1305 vector that harbored a 3XFLAG tag. This construct was used for transgenic complementation of mos12-1 snc1 npr1. The construct used for MOS12-GFP analysis was created in the same way, but ligated to the modified pCAMBIA1305 vector that contained a GFP tag. The primers used for the amplification of the fragment are as follows: MOS12-F, 5′-CGCGGATCCAGAGGAGAGCGCCATTTGAG-3′; MOS12-R, 5′-ACGCACGCGTCGACATGGTGCCTACGACGGTCTTTC-3′.
Total protein extraction
For total protein extraction, 0.1 g of leaf tissue from 4-week-old soil-grown plants was harvested and ground using liquid nitrogen. The samples were homogenized in extraction buffer (100 mM Tris–HCl, pH 8, 0.1% SDS, 2%β-mercaptoethanol). After 5 min of centrifugation at 16100 g in a microcentrifuge tube, the supernatant was transferred to a new microfuge tube and boiled for 5 min at 100°C after the addition of 4X SDS loading buffer. Samples were analyzed by western blot analysis.
Nuclear extraction and immunoprecipitation
Approximately 20 g of seedling from a complementing MOS4-HA transgenic line in the mos4-1 background was frozen in liquid nitrogen (Palma et al., 2007), ground to a fine powder and homogenized in lysis buffer (20 mm Tris-HCl, pH 7.4, 25% glycerol, 20 mm KCl, 2 mm EDTA, 2.5 mm MgCl2, 250 mm sucrose) at 4°C. The homogenate was sequentially filtered through a 100- and 30-μm nylon mesh. The nuclei were pelleted by centrifugation at 1500 g for 10 min and washed three times with nuclei resuspension buffer (NRBT; 20 mm Tris–HCl, pH 7.4, 25% glycerol, 2.5 mm MgCl2, 0.2% Triton X-100) at 4°C. The nuclei were then resuspended in 2 ml ice-cold buffer NE-2 (20 mm HEPES-KOH, pH 7.9, 2.5 mm MgCl2, 250 mm NaCl, 20% glycerol, 0.2% Triton X-100, 0.2 mm EDTA, 1 mm DTT and protease inhibitor cocktail from Sigma-Aldrich, http://www.sigmaaldrich.com) and then ultracentrifugated at 35 000 g for 15 min, which helped to break the nuclear membrane. The pellet was resuspended in the same buffer and subjected to sonication for 4 min with 5-sec on and 10-sec off intervals to completely release nuclear proteins after centrifugation at 13 200 rpm for 30 min. The supernatant was separated into two equal volume samples. One sample was mixed with 50 μl of anti-GFP and the other with 50 μl of anti-HA MicroBeads (Miltenyi Biotec, http://www.miltenyibiotec.com). The anti-GFP treatment served as a negative control. After incubation at 4°C for 1 h, the MicroBead-bound target protein was magnetically precipitated on columns according to the manufacturer’s instructions (μMACs; Miltenyi Biotec). The columns were then washed eight times with buffer NE-3 (20 mm HEPES-KOH, pH 7.9, 2.5 mm MgCl2, 150 mm NaCl, 20% glycerol, 0.2% Triton X-100, 0.2 mm EDTA, 1 mm DTT, with protease inhibitors) before proteins were eluted with 60 μl of 95°C pre-heated 1XSDS loading buffer. The samples were subsequently analyzed by western blot. MOS12 antibody was generated against its purified full-length protein from rabbit. The elution fraction is about 12.5-fold more concentrated as the input fraction. For the western blot probed with antibodies against HA, AtCDC5 (Palma et al., 2007) and HDA19 (Zhu et al., 2010), 25 μl of elution sample and 15 μl of input sample were loaded. For the anti-MOS12 western blot, 20 μl of elution sample and 2 μl of input sample were loaded.
Dr Walter Gassmann is thanked for the Pst avrHopA1 strain for testing RPS6-mediated immunity. Ms Yu Ti Cheng, Dr EunKyoung Lee and Dr Fred Sack are thanked for their kind help with confocal microscopy. Ms Kaeli Johnson and Ms Virginia Woloshen are thanked for their careful reading of the manuscript. Dr Shao-Lun Liu is thanked for helping with phylogenetic analysis. We are grateful for financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery program and the 973 program of the Chinese Ministry of Science and Technology, grant number 2011CB10070.