Two putative RNA-binding proteins function with unequal genetic redundancy in the MOS4-associated complex.

The MOS4-associated complex (MAC) is a highly conserved nuclear protein complex associated with the spliceosome. We recently purified the MAC from Arabidopsis (Arabidopsis thaliana) nuclei, identified its potential components by mass spectrometry, and showed that at least five core proteins in the MAC are required for defense responses in plants. Here, we report the characterization of a putative RNA-binding protein identified in the MAC named MAC5A and its close homolog MAC5B. We confirmed that MAC5A is a component of the MAC through coimmunoprecipitation with the previously described MAC protein CELL DIVISION CYCLE5 from Arabidopsis. In addition, like all other characterized MAC proteins, MAC5A fused to the Green Fluorescent Protein localizes to the nucleus. Double mutant analysis revealed that MAC5A and MAC5B are unequally redundant and that a double mac5a mac5b mutant results in lethality. Probably due to this partial redundancy, mac5a and mac5b single mutants do not exhibit enhanced susceptibility to virulent or avirulent pathogen infection. However, like other MAC mutations, mac5a-1 partially suppresses the autoimmune phenotypes of suppressor of npr1-1, constitutive1 (snc1), a gain-of-function mutant that expresses a deregulated Resistance protein. Our results suggest that MAC5A is a component of the MAC that contributes to snc1- mediated autoimmunity.

Higher plants have evolved a sophisticated innate immune system against microbial pathogen infection. Pattern recognition receptors and Resistance (R) proteins represent two major classes of receptors in plants that recognize the presence of pathogens and trigger cell-autonomous defense responses (Jones and Dangl, 2006). Many pattern recognition receptors are transmembrane receptor-like kinases that contain extracellular leucine-rich repeats (LRRs) and are involved in the perception of nonself epitopes common to whole groups of microbes called microbe-associated molecular patterns (Zipfel, 2009). Most cloned R genes encode intracellular proteins containing a central nucleotide-binding (NB) site and C-terminal LRRs and are involved in the recognition of effector proteins secreted by pathogens during infection. The activation of NB-LRRs is often associated with the onset of a cell death program known as the hypersensitive response, in which the plant cell kills itself to restrict pathogen growth. Most mutants with deregulated NB-LRR proteins exhibit seedling lethality, suggesting that NB-LRRs are under tight negative regulation to avoid unnecessary cell death (Lukasik and Takken, 2009). SUPPRESSOR OF NPR1-1, CONSTITUTIVE1 (SNC1) encodes a NB-LRR with an N-terminal Toll/Inter-leukin1 Receptor (TIR) domain (Li et al., 2001b;Zhang et al., 2003). The unique gain-of-function mutant snc1 exhibits constitutive defense responses in the absence of cell death, representing an autoimmune model amenable to genetic analysis. A forward genetic screen for suppressors of snc1 revealed that snc1-mediated defense includes components involved in nucleocytoplasmic partitioning Cheng et al., 2009), protein modification (Goritschnig et al., 2007(Goritschnig et al., , 2008, and RNA processing Palma et al., 2007). Importantly, like most other TIR-NB-LRRs, snc1 signaling requires the function of PHYTOALEXIN DEFICIENT4 and EN-HANCED DISEASE SUSCEPTIBILITY1 (EDS1; Li et al., 2001b;Zhang et al., 2003).
MODIFIER OF SNC1, 4 (MOS4) is the founding member of the MOS4-associated complex (MAC; Palma et al., 2007), a highly conserved spliceosome-associated com-1 This work was supported by the Natural Sciences and Engineering Research Council of Canada and the Canadian Foundation for Innovation (Discovery grants to X.L.) and by the National Science Foundation of the People's Republic of China (grant to Y.Z.). J.M. was supported by a Natural Sciences and Engineering Research Council postgraduate scholarship for doctoral students.
2 Present address: The Sainsbury Laboratory, Colney Lane, Norwich NR4 7UH, United Kingdom. * Corresponding author; e-mail xinli@interchange.ubc.ca. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Xin Li (xinli@interchange.ubc.ca). [C] Some figures in this article are displayed in color online but in black and white in the print edition. [W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription.
Here, we characterize two unequally redundant putative RNA-binding proteins, MAC5A and MAC5B, with sequence similarity to the human RNA-binding motif protein RBM22. MAC5A was previously isolated as a component of the MAC (Monaghan et al., 2009) but has otherwise not yet been studied in plants. We show that MAC5A localizes to the nucleus and interacts with AtCDC5 in planta, confirming its association with the MAC. In addition, we show that MAC5A and its close homolog MAC5B are partially redundant in a dosage-dependent manner and that a double mac5a mac5b mutant is lethal. Although single mac5a and mac5b mutants do not exhibit obvious enhanced susceptibility to pathogen infection, we found that mac5a-1 suppresses snc1-associated phenotypes. Overall, our results indicate that MAC5A is a component of the MAC that functions in the snc1 signaling pathway.

Isolation of mac5a, mac5b, and mac5c T-DNA Insertion Mutants
We used the MAC5A protein sequence as a query in BLAST and identified two additional proteins with significant sequence similarity to MAC5A (At1g07360) encoded in the Arabidopsis genome. We named these proteins MAC5B (At2g29580) and MAC5C (At5g07060). MAC5A and MAC5B are proteins of approximately 480 amino acids in length that are 82% identical and contain a CCCH-type zinc-finger domain and an RNA recognition motif (RRM; Fig. 1A; Supplemental Fig. S1). Conversely, MAC5C is a truncated protein of 363 amino acids that contains only a zinc-finger domain and no RRM ( Fig. 1A; Supplemental Fig. S1). The phylogenetic relationship between these proteins indicates that MAC5A and MAC5B are more closely related to each other than to MAC5C (Addepalli and Hunt, 2008;Wang et al., 2008). In addition, according to publicly available microarray data (Toufighi et al., 2005;Winter et al., 2007), MAC5A and MAC5B are expressed in similar tissue types, although MAC5A is expressed at a much higher level (Supplemental Fig. S2). Conversely, MAC5C is expressed at very low levels in dry seeds, senescent leaves, and floral organs but not at all in any other tissues (Toufighi et al., 2005;Winter et al., 2007;Supplemental Fig. S2). MAC5A, MAC5B, and MAC5C share homology with the human protein RBM22/ hECM2/fSAP47 (42%-50% identity at the amino acid level) and share very weak homology with the yeast protein Ecm2p/Slt11p (13%-16% identity; Supplemental Fig. S1). These proteins have been repeatedly isolated as components of the NTC/MAC in several eukaryotes (Ohi et al., 2002;Deckert et al., 2006;Gavin et al., 2006;Bessonov et al., 2008;Herold et al., 2009;Monaghan et al., 2009).
To study the biological functions of MAC5A, MAC5B, and MAC5C, we obtained T-DNA insertion alleles from Arabidopsis seed centers. Salk_132881 (mac5a-1) carries an insertion in the second exon of MAC5A, whereas Salk_142085 (mac5a-2) and Salk_ 072670 (mac5a-3) carry insertions in the first intron of MAC5A (Fig. 1A). Semiquantitative reverse transcription (RT)-PCR indicated that all alleles were genuine mac5a knockout mutants, as no MAC5A transcript could be detected compared with ecotype Columbia (Col-0; Supplemental Fig. S3B). The leaves of these mac5a mutants appeared slightly serrated, twisty, and often exhibited elongated petioles ( Fig. 1B; Supplemental Fig. S3A). In addition, mac5a mutants displayed shortened roots compared with Col-0 plants (Supplemental Fig. S3C), were early flowering (Supplemental Fig. S3A), and had reduced seed yield (data not shown). Only plants homozygous for the mac5a mutations displayed the same morphological phenotypes, indicating that the alleles are recessive and that the phenotypes cosegregate with the mutations. Importantly, F1 progeny from crosses between these mutants did not exhibit phenotypic complementation (Supplemental Fig. S3A), confirming that they are allelic.
GK419F11 carries an insertion in the first exon of MAC5B (Fig. 1A). This particular T-DNA insertion line is annotated as also being in the promoter region of PCNA2, a locus that neighbors MAC5B in the opposite orientation on chromosome II (Supplemental Fig.  S4A). Because of this, we examined the expression levels of both genes in GK419F11 compared with Col-0 using semiquantitative RT-PCR. We found that expression of MAC5B was strongly reduced in this line, whereas the expression of PCNA2 was not affected (Supplemental Fig. S4B). This indicated that GK419F11 is a null allele of MAC5B, so we named it mac5b-1. Plants homozygous for the insertion did not exhibit any aberrant phenotypes and appeared morphologically similar to wild-type Col-0 plants (Fig. 1B). Although our in silico analysis indicated that MAC5C is barely expressed in Arabidopsis, we sought to test the biological function of this locus using a knockout mutant as well. For this, we obtained WiscDsLox262F09, which carries an insertion in the fourth intron of MAC5C, very close to an intron-exon junction (Fig.  1A). Like mac5b-1 mutant plants, these mac5c-1 plants also appeared indistinguishable from Col-0 ( Fig. 1B).

MAC5A and MAC5B Are Unequally Redundant
Because of the sequence similarity between MAC5A, MAC5B, and MAC5C, we were interested to test for possible genetic redundancy between these loci. To do this, we crossed homozygous mutants together to create mac5a-1 mac5b-1, mac5a-1 mac5c-1, and mac5b-1 mac5c-1 double mutants. The mac5a-1 mac5c-1 double mutants appeared morphologically similar to mac5a-1 single mutants (Supplemental Fig. S5A), and the mac5b-1 mac5c-1 double mutants appeared morphologically similar to Col-0 wild-type plants (Supplemental Fig. S5B). These data suggest that there are no redundant functions shared between MAC5C and MAC5A or MAC5B. In contrast, MAC5A and MAC5B seem to perform some redundant functions in plants, as a double mac5a-1 mac5b-1 mutant is lethal. Out of 127 randomly chosen progeny from a parent that was homozygous for mac5b-1 but heterozygous for mac5a-1, 52 were wild type for MAC5A, 75 were heterozygous for mac5a-1, and none was homozygous for mac5a-1, indicating a 1:2:0 ratio rather than a normal 1:2:1 ratio. A x 2 test using these values indicates that a combination of mac5a-1 and mac5b-1 is lethal (x 2 = 3.263; P = 0.07). From another segregating population in which the parent was heterozygous for both mac5a-1 and mac5b-1, we recovered very few plants (typically one or two out of approximately 100 plants) that were homozygous for mac5a-1 and heterozygous for mac5b-1. These plants, in which there was only one functional copy of MAC5B and no functional copies of MAC5A, displayed severe developmental defects, including dwarfism, delayed growth, abnormal floral organs, and sterility ( Fig. 2A). Terminal outgrowths known as enations were also observed extending from the leaf margins of these plants (Fig. 2B). To further investigate the redundant roles of MAC5A and MAC5B, we tested if overexpression of MAC5B can compensate for the loss of MAC5A in mac5a-1 mutants. For this, we stably expressed MAC5B under the control of the constitutive cauliflower mosaic virus P35S promoter in homozygous mac5a-1 plants by Agrobacterium tumefaciens-mediated transformation. Analysis of transgenic plants revealed that expression of P35S-MAC5B was indeed able to complement mac5a-1 ( Fig.  2A). Overexpression of MAC5B in the transgenics was confirmed by semiquantitative RT-PCR (Fig. 2C). These results further supported redundant functions between MAC5A and MAC5B and suggested that the major difference between these loci is linked to their steady-state expression levels. Thus, because a double mac5a-1 mac5b-1 mutant is lethal and overexpression of MAC5B can compensate for the loss of MAC5A, we conclude that MAC5A and MAC5B are unequally redundant, as defined by Briggs et al. (2006), and that MAC5A is the dominant contributor of the pair. mac5a-1 Suppresses snc1 Autoimmunity MOS4, AtCDC5, and MAC3A/3B are required for snc1-mediated immunity, as deduced from double mutant analysis between snc1 and mos4, Atcdc5, or mac3a mac3b mutants (Palma et al., 2007;Monaghan et al., 2009). To test if MAC5A, MAC5B, or MAC5C is likewise required for snc1 signaling, we created snc1 mac5a-1, snc1 mac5b-1, and snc1 mac5c-1 double mutants and conducted suppression analysis of snc1related phenotypes. The loss of MAC5A (Fig. 3), but not MAC5B or MAC5C (Supplemental Fig. S6), sup- aa, Amino acids. The phylogenetic relationship between the encoded proteins is indicated to the left, as predicted by Addepalli and Hunt (2008) and Wang et al. (2008). B, Morphology of homozygous mac5a-1, mac5b-1, and mac5c-1 mutants compared with Col-0. Soil-grown plants were photographed 3 weeks after germination. Bar = 1 cm. [See online article for color version of this figure.] pressed snc1-associated dwarfism, as snc1 mac5a-1 plants were more than twice the size of snc1 (Fig.  3A). Heightened resistance to Hyaloperonospora arabidopsidis (H.a.) isolate Noco2 and Pseudomonas syringae pathovar tomato (P.s.t.) strain DC3000 was also suppressed in snc1 mac5a-1, as indicated by quantification of pathogen growth in these genotypes compared with the Col-0 wild type. As shown in Figure 3B, whereas H.a. Noco2 was unable to colonize snc1 plants, snc1 mac5a-1 mutants were as susceptible as Col-0 to this pathogen. Infection with P.s.t. DC3000 indicated that snc1 mac5a-1 plants sustain 100-fold higher bacterial growth over 3 d compared with snc1 (Fig. 3C). Furthermore, quantitative RT-PCR showed that the constitutive expression of PR-1 and PR-2 in snc1 is partially suppressed by mac5a-1 (Fig. 3D). Importantly, accumulation of the SNC1 transcript did not differ between the snc1 and snc1 mac5a-1 mutants (Fig. 3D), indicating that altered SNC1 expression is not the likely cause for the observed suppression.
We previously demonstrated that MAC5A and MAC5B are unequally redundant and that MAC5A is the dominant contributor. Thus, it is likely that the reason mac5b-1 was unable to suppress snc1 autoimmunity was simply due to normal expression of MAC5A in the double mutants. To investigate the role of MAC5B in snc1 autoimmunity, we stably expressed P35S-MAC5B in snc1 mac5a-1 plants and tested for restoration of the snc1 phenotype. As a control, we also stably expressed PMAC5A-MAC5A in snc1 mac5a-1. Out of 15 P35S-MAC5B transgenics, nine restored snc1 morphology (Supplemental Fig. S7). Similarly, out of 12 PMAC5A-MAC5A transgenics, nine restored snc1 morphology (Supplemental Fig. S7). These results indicate that MAC5B can function in the snc1 pathway and further demonstrate unequal genetic redundancy between MAC5A and MAC5B in Arabidopsis. Mutations in MOS4, AtCDC5, PRL1, and MAC3A/3B lead to enhanced susceptibility to pathogen infection (Palma et al., 2007;Monaghan et al., 2009). To test if any of the three MAC5 homologs are necessary for the full expression of basal defense responses in Arabidopsis, mac5a, mac5b, and mac5c mutants were challenged with P.s.t. DC3000 and bacterial growth was assayed after 3 d. We observed similar bacterial growth in Col-0 and the mutants ( Fig. 4A; Supplemental Fig. S8), whereas the well-known enhanced susceptibility mutant npr1-1 (Cao et al., 1994) harbored an over 10-fold higher titer of bacteria in all experiments. In addition, when we infected plants with H.a. Noco2, we consistently observed slightly higher oomycete colonization of the mac5a mutants compared with Col-0; however, this difference could not be repeatedly supported by statistical analysis (Fig. 4B).
To test if R protein-mediated defenses other than snc1 autoimmunity require MAC5A or MAC5B, we also infected plants with pathogens that express avirulence effectors specifically recognized by R proteins in Col-0. P.s.t. avrRps4 and P.s.t. avrPphB (recognized by RPS4 and RPS5, respectively) grew to similar levels in Col-0 and the mac5a and mac5b mutants (Fig. 4, C and D). In addition, the H.a. isolates Emwa1 and Cala2, expressing avirulence effectors recognized by RPP4 and RPP2, respectively, grew to similar levels in Col-0 and mac5a-1 (data not shown), and both genotypes were able to mount a hypersensitive response at infection sites, as indicated by lactophenol trypan blue staining on leaf tissue 7 d after inoculation (Supplemental Fig. S9). Together, these data reveal that although MAC5A is required for snc1-mediated defense, MAC5A and MAC5B may not be required for other R protein-mediated pathways or basal defense in plants. However, the involvement of MAC5A or MAC5B in these pathways may be masked by the partial redundancy between the two loci.

MAC5A Localizes to the Nucleus and Associates with AtCDC5 in Planta
To confirm that MAC5A is a MAC component, we tested whether MAC5A localizes to the nucleus and if it is capable of associating with the MAC protein AtCDC5 in a coimmunoprecipitation assay. For this, we stably transformed either PMAC5A-MAC5A-GFP or PMAC5A-MAC5A-HA into mac5a-1 plants by Agrobacterium-mediated transformation. The majority of transgenic lines (24 of 25 independent GFP lines and 21 of 22 independent hemagglutinin [HA] lines) complemented mac5a-1 phenotypes such as morphology (Fig. 5A) and root length (Fig. 5B). Because these fusion proteins complemented mac5a-1 phenotypes, they should function similarly to endogenous MAC5A. We confirmed that the fusion proteins were expressed in vivo by extracting total protein from representative transgenic lines followed by western-blot analysis using antibodies against GFP or HA (data not shown).
All MAC5A-GFP complementing lines showed clear nuclear localization when viewed with a fluorescence microscope. Root cells from one representative line are shown in Figure 5C, but nuclear localization was also observed in other tissues, including leaves (data not shown). This localization was not unexpected, as the MAC5A homolog in zebrafish, RBM22, also localizes to the nucleus (Montaville et al., 2006), as do the MAC core proteins MOS4, AtCDC5, PRL1, and MAC3A/3B. To test if MAC5A associates with AtCDC5, we isolated total nuclear protein and immunoprecipitated MAC5A-HA using anti-HA microbeads. Western-blot analysis using an anti-AtCDC5 antibody revealed that AtCDC5 was present only in the eluted fraction from transgenic plants and not from control Col-0 plants (Fig. 6). Association between MAC5A and AtCDC5 has been shown in other eukaryotes as well (McDonald et al., 1999;Ohi et al., 2002;Deckert et al., 2006;Gavin et al., 2006;Bessonov et al., 2008). These data indicate that MAC5A localizes to the nucleus and associates with AtCDC5 in planta. Thus, we conclude that MAC5A is indeed a member of the MAC. Values represent averages of two replicates of 20 seedlings each 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference from group a (P , 0.008). This experiment was repeated three times with similar results. FW, Fresh weight. C, Growth of P.s.t. DC3000 at 0 and 3 d after infection. Values represent averages of four replicates 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference from group a (P , 0.0008). This experiment was repeated six times with similar results. cfu, Colony-forming units. D, Steady-state expression analysis of PR-1, PR-2, and SNC1 relative to Col-0 and normalized against ACTIN1 using quantitative RT-PCR. Values represent averages of three experimental replicates 6 SD. This experiment was repeated using cDNA isolated from five independent trials with similar results. [See online article for color version of this figure.]

Analysis of Constitutive Splicing in Higher Order MAC Mutants
Because of the close association between the MAC and the spliceosome, we previously tested if general splicing was impaired in the MAC mutants. For this, we monitored the presence of several known alternative transcripts in MAC mutants compared with Col-0 (Palma et al., 2007). We did not observe obvious differences in the presence of U1-70K, AtSRp34/SR1, or AtSRp30 alternative transcript variants (Palma et al., 2007), leading us to conclude that the MAC is not required for general splicing in Arabidopsis. However, although we have been unable to show aberrations in general/constitutive splicing, the MAC as a whole appears to be essential for plant viability, as all higher order MAC mutants so far described (mos4-1 Atcdc5-1, mos4-1 prl1-1, mos4-1 mac3a mac3b, and mac3a mac3b prl1-1) result in lethality (Palma et al., 2007;Monaghan et al., 2009). Probably because of the partial redundancy between MAC5A and MAC5B, we were able to successfully isolate double mutants when we crossed mos4-1, Atcdc5-1, prl1-2, or mac3a mac3b with mac5a-1. The morphological phenotypes of the double mutants looked like a combination of phenotypes between the respective loci (Fig. 7A). The most striking phenotype was observed with mac5a-1 prl1-2 double mutant plants, which exhibited severe developmental defects and often displayed leaf enations similar to the mac5a-1/mac5a-1 MAC5B/mac5b-1 mutant. Interestingly, when we infected plants with P.s.t. DC3000 and monitored growth 3 d later, we did not observe a further increase in bacterial titer in any of the combination mutants (Supplemental Fig. S10; data not shown), indicating that the additional loss of MAC5A does not further impair basal defenses. When we tested the fidelity of Values represent averages of four replicates 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference from group a (P , 0.0005). This experiment was repeated six times with similar results. cfu, Colony-forming units. B, Growth of H.a. isolate Noco2 at 7 d after infection with 2.5 3 10 4 spores mL 21 water. Values represent averages of four replicates of 20 seedlings each 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference compared with group a (P , 0.005). This experiment was repeated three times with similar results. FW, Fresh weight. C and D, Growth of the bacterial pathogens P.s.t. avrRps4 and P.s.t. avrPphB at 0 and 3 d postinoculation. Values represent averages of four replicates 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference from group a (P , 0.0001). These experiments were repeated six times with similar results. the splicing machinery in these higher order MAC mutants, we again did not observe any changes in the presence of U1-70K, AtSRp34/SR1, or AtSRp30 alternative transcript variants (Fig. 7B). We also did not observe any changes in these transcripts in the mac5a-1/mac5a-1 MAC5B/mac5b-1 mutant (data not shown). These data indicate that the spliceosome is functional in MAC single and higher order mutants and further demonstrate that, individually, MOS4, AtCDC5, PRL1, MAC3A/3B, and MAC5A/5B do not contribute key functions to general splicing in plants, even though the MAC is closely associated with proteins that form the spliceosome. It is important to keep in mind, however, that our analysis is not stringent enough to observe even severalfold differences in the abundance of splice variants, so we are not able to address splicing efficiency in the MAC mutants at this time. We also cannot rule out the possible involvement of the MAC in specific splicing events such as those that may be important during the plant immune response.

DISCUSSION
The MAC is a spliceosome-associated protein complex homologous to the well-studied NTC in yeast and the CDC5L complex in human (Palma et al., 2007;Monaghan et al., 2009). It contains proteins with diverse functions, including the scaffolding protein MOS4, the Myb transcription factor AtCDC5, the WD-40 protein PRL1, the E3 ubiquitin ligases MAC3A and MAC3B, as well as several nucleic acid-binding proteins and small nuclear ribonucleoprotein (snRNP) subunits that are predicted to be integral components of the spliceosome (Monaghan et al., 2009). Here, we present evidence suggesting that the putative RNAbinding protein MAC5A is unequally redundant with its close homolog MAC5B and functions in snc1mediated autoimmunity as a component of the MAC.
MAC5A and MAC5B are predicted to be RNAbinding proteins by virtue of containing both an RRM and a CCCH-type zinc-finger motif. Although the intrinsic ability of MAC5A/5B to bind RNA species has yet to be shown experimentally, proteins containing CCCH zinc fingers or RRMs have been demonstrated to bind RNA in vitro (Jessen et al., 1991;Burd and Dreyfuss, 1994;Wang et al., 2008). The most abundant RNA-binding domain in eukaryotes is the RRM (Lorković and Barta, 2002), often found in proteins also containing CCCH motifs, such as in MAC5A/5B. There are 196 RRM-containing proteins encoded in the Arabidopsis genome, including several predicted poly(A)-binding proteins, Arg-rich (SR) proteins, and snRNPs (Lorković and Barta, 2002). Importantly, it has been demonstrated that the RRM is necessary and sufficient for RNA binding (Jessen et al., 1991;Burd and Dreyfuss, 1994). In addition to Figure 5. Subcellular localization of MAC5A-GFP. A, Full-length genomic MAC5A-GFP driven by its native promoter complements mac5a-1 morphology. Soil-grown plants were photographed 4 weeks after germination. Bar = 1 cm. B, Root length assay demonstrating complementation of mac5a-1 by the MAC5A-GFP transgene. Plants were grown on MS medium supplemented with 0.5% Suc and 0.3% phytagel for 8 d. Values represent means of 10 replicates 6 SD. Statistically significant groups were calculated using multiple unpaired Student's t tests comparing the means of all samples against Col-0; a indicates no significant difference compared with Col-0, and b indicates a statistically significant difference from group a (P , 0.0001). This experiment was repeated three times with similar results. C, MAC5A-GFP localizes to the nucleus. Confocal microscopy was used to examine the localization of MAC5A-GFP in the transgenic line shown in A. Root cells from a representative line are shown. 4,6-Diamidino-2-phenylindole (DAPI) was used as a control for nuclear localization. DIC, Differential interference contrast.
the RRM, there are several other RNA-binding domains, such as the CCCH zinc-finger motif. This family is also fairly large in Arabidopsis, with 68 proteins encoded in the genome (Wang et al., 2008). Whereas most zinc fingers are involved in DNA binding, several zinc-finger-containing CCCH motifs have been shown to function posttranscriptionally and to have RNA-binding capability (Wang et al., 2008). For example, the Arabidopsis protein ENHANCER OF AG-4, 1 contains six tandem CCCH motifs and binds RNA in vitro (Li et al., 2001a). Likewise, the Arabi-dopsis cleavage and polyadenylation specificity factor AtCPSF30 contains three tandem CCCH motifs and has also been shown to bind RNA (Delaney et al., 2006;Wang et al., 2008). Interestingly, a recent study demonstrated that five CCCH zinc-finger proteins in Arabidopsis, representing proteins from diverse subfamilies, are capable of degrading RNA in vitro, uncovering the intriguing possibility that ribonuclease activity may be a common property of CCCH proteins in Arabidopsis (Addepalli and Hunt, 2008).
It is not hard to imagine roles for RNA-binding proteins and/or ribonucleases in plant defense and snc1 autoimmunity. A large number of genes are differentially regulated following the perception of pathogens, and this massive change in gene expression must be correlated with subsequent RNA processing and/or degradation events. A relevant example is the Gly-rich RRM-containing protein AtGRP7, which was recently shown to be targeted by HopU1, a Pseudomonas type III effector with mono-ADP-ribosyltransferase activity, to suppress the plant immune response and increase virulence (Fu et al., 2007). Loss-of-function Atgrp7 mutants were shown to display enhanced susceptibility to infection by Pseudomonas in the same study. In addition, MOS2, a protein containing RNA-binding G-patch and KOW motifs, is likewise required for innate immunity in Arabidopsis , although the precise Figure 6. MAC5A-HA and AtCDC5 coimmunoprecipitate in planta. Total nuclear extracts were isolated from a complementing mac5a-1 transgenic line expressing PMAC5A-MAC5A-HA (+) and Col-0 (2). MAC5A-HA was immunoprecipitated using anti-HA microbeads. MAC5A-HA and AtCDC5 were detected in the eluted fractions (203 concentrated) by western-blot analysis using antibodies against HA or AtCDC5. Figure 7. Analysis of constitutive splicing in higher order MAC mutants. A, Morphology of soil-grown plants photographed 4 weeks after planting. Bar = 1 cm. B, The presence of alternative transcript variants for the genes UI-70K, AtSRp34/SR1, and AtSRp30 was monitored in Col-0 compared with the MAC mutants mos4-1, Atcdc5-1, prl1-2, mac3a mac3b, mac5a-1, and the higher order mutants. An equal amount of RNA from all genotypes was used to make cDNA for subsequent analysis by semiquantitative RT-PCR. TUBULIN was included as a control. This experiment was repeated four times from independent trials. The schematic to the right represents the nature of the alternative transcript variants as adapted from Palma et al. (2007). [See online article for color version of this figure.] function of this protein is unclear. Our finding that MAC5A is partly required for snc1-mediated autoimmunity points to a possible role for this RNA-binding protein in plant defense. However, we did not observe enhanced susceptibility in mac5a or mac5b single mutants when we infected plants with a variety of pathogens, possibly complicated by the partial redundancy between MAC5A and MAC5B. The lethality of the mac5a-1 mac5b-1 double mutant prevented us from testing whether plants lacking both MAC5A and MAC5B exhibit enhanced disease susceptibility like the other MAC mutants characterized to date.
Importantly, whereas the MAC core components MOS4, AtCDC5, PRL1, and MAC3A/3B are not encoded by essential genes in Arabidopsis, both MAC5A and MAC5B are required for plant viability. Because of the physical association between the MAC and components of the spliceosome, it is very likely that the essential functions employed by MAC5A and MAC5B have something to do with RNA processing. That said, we have not yet been able to directly link the function of the MAC to splicing. We can only speculate that this association has to do with transcriptional coregulation or alternative splicing of currently unknown gene targets. The presence of transcript variants for three known alternatively spliced genes, U1-70K, AtSRp34/SR1, and AtSRp30, was not affected in mac5a-1, mac5a-1/mac5a-1 MAC5B/mac5b-1, or any other MAC mutants, broadly suggesting that the spliceosome is functional despite these lesions. And yet, the loss of any two MAC components in combination results in synthetic lethality, indicating that the MAC as a whole is required for an essential function, such as splicing or spliceosome assembly, as in yeast (Chan et al., 2003). In this respect, the identification of MAC substrates and downstream signaling components is of particular interest. Uncovering what genes are regulated by AtCDC5, what proteins are ubiquitinated by MAC3A/3B, and what RNA species are bound by MAC5A/5B will truly allow us to elucidate the molecular function of the MAC. A major hurdle, however, will be deciphering which downstream components are specific to the plant defense response. It is clear that MAC proteins are employed by several signaling networks, as indicated by the pleiotropic phenotypes observed for MAC mutants described to date. All MAC mutants, to one extent or another, show defects in flowering time, leaf development, fertility, root length, hormone signaling, sugar signaling, and innate immunity (Németh et al., 1998;Palma et al., 2007;Monaghan et al., 2009;J. Monaghan and X. Li, unpublished data). Based on these observations, it is possible that the MAC acts as a key regulatory node utilized by many signaling pathways in plants. A joint effort to study the molecular function of this complex across kingdoms could shed light on the role of the MAC/NTC in signal transduction relay and may lead to a better understanding of how plants utilize this complex to achieve immunity.

Plant Growth and Pathology Assays
Arabidopsis (Arabidopsis thaliana) plants were grown on soil in a 16-hlight/8-h-dark regime or on Murashige and Skoog (MS) medium supplemented with 0.5% Suc and 0.3% phytagel or 0.8% agar, depending on the experiment. Bacterial and oomycete infections were performed as described (Li et al., 2001b). Briefly, bacterial pathogens were inoculated on the abaxial leaf surfaces of 4-week-old plants using a needless syringe. Leaf discs (with an area of 0.38 cm 2 ) were collected on the day of infection (day 0) and 3 d later (day 3) from different plants. Hyaloperonospora arabidopsidis isolates were spray inoculated onto adaxial leaf surfaces of 2-week-old seedlings and stained using lactophenol trypan blue 7 d later using a protocol described by Koch and Slusarenko (1990).

Nuclear Protein Extraction and Immunoprecipitation
Approximately 15 g of leaf tissue from complementing mac5a-1 plants expressing PMAC5A-MAC5A-HA was used to isolate the nuclear protein fraction with a procedure described by Monaghan et al. (2009). Immunoprecipitation was carried out using anti-HA microbeads (Miltenyi Biotec) as described (Monaghan et al., 2009). The eluted fraction was loaded on a 12% SDS-PAGE gel followed by western-blot analysis using anti-HA (Roche) or anti-AtCDC5 (Palma et al., 2007) antibodies.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Multiple sequence alignment of MAC5A homologs in Arabidopsis, human, and budding yeast.
Supplemental Figure S4. Expression analysis of PCNA2 and MAC5B in the mac5b-1 mutant.
Supplemental Figure S10. Growth of P.s.t. DC3000 in higher order MAC mutants.
Supplemental Table S1. Summary of phenotypes reported for MAC mutants in Arabidopsis.