Arginine methylation of SM-LIKE PROTEIN 4 antagonistically affects alternative splicing during Arabidopsis stress responses

Abstract

Arabidopsis (Arabidopsis thaliana) PROTEIN ARGININE METHYLTRANSFERASE5 (PRMT5) post-translationally modifies RNA-binding proteins by arginine (R) methylation. However, the impact of this modification on the regulation of RNA processing is largely unknown. We used the spliceosome component, SM-LIKE PROTEIN 4 (LSM4), as a paradigm to study the role of R-methylation in RNA processing. We found that LSM4 regulates alternative splicing (AS) of a suite of its in vivo targets identified here. The lsm4 and prmt5 mutants show a considerable overlap of genes with altered AS raising the possibility that splicing of those genes could be regulated by PRMT5-dependent LSM4 methylation. Indeed, LSM4 methylation impacts AS, particularly of genes linked with stress response. Wild-type LSM4 and an unmethylable version complement the lsm4-1 mutant, suggesting that methylation is not critical for growth in normal environments. However, LSM4 methylation increases with abscisic acid and is necessary for plants to grow under abiotic stress. Conversely, bacterial infection reduces LSM4 methylation, and plants that express unmethylable-LSM4 are more resistant to Pseudomonas than those expressing wild-type LSM4. This tolerance correlates with decreased intron retention of immune-response genes upon infection. Taken together, this provides direct evidence that R-methylation adjusts LSM4 function on pre-mRNA splicing in an antagonistic manner in response to biotic and abiotic stress.

IN A NUTSHELL

Background: Plants are constantly subjected to a variety of environmental fluctuations, ranging from alterations in ambient temperature to infection by pathogens. In response to these stresses, plant cells trigger a complex reprogramming of gene expression, altering not only growth and development but also eliciting defense mechanisms. Several levels of gene regulation are involved in plant stress responses. These include transcriptional regulation, which determines which genes are activated; post-transcriptional regulation, which controls if mRNAs undergo alternative splicing; translational regulation, which controls when mRNAs are translated into proteins; and post-translational regulation, which controls whether these proteins suffer additional modifications that influence their activity. Consequently, the interaction of the different molecular programs coordinating gene expression is essential for plant development.

Question: Previous studies established that in Arabidopsis plants, the PRMT5 methyltransferase protein methylates the arginine residues of proteins that are involved in the regulation of alternative splicing. To determine the importance of methylation in alternative splicing regulation, especially in response to stress, we employed the SM-LIKE PROTEIN 4 (LSM4) splicing factor as a model.

Findings: We show that loss of LSM4 leads to widespread missplicing, but its methylation has a restricted role in splicing under normal growth. On the other hand, LSM4 methylation responds to environmental cues associated with stress, showing antagonic responses under both biotic and abiotic conditions. Bacterial infection leads to decreased LSM4 methylation, influencing the splicing of genes associated with plant immunity to enhance resistance. In contrast, salinity increases LSM4 methylation, crucial for the correct splicing of genes involved in abiotic stress responses.

Next steps: Our work shows how post-translational modification changes RNA splicing to allow plants to acclimate to the environment. The interaction of diverse protein modifications, converging to efficiently control the RNA splicing process, has the potential to reveal key molecular mechanisms essential for plant adaptation.

Introduction

The ability of plants to acclimatize and maintain high yields under stressful environmental conditions is a topic of heightened concern due to climate change and increasing demands on agriculture. Plant stress acclimation involves multiple layers of post-transcriptional (Hernando et al. 2017; Dikaya et al. 2021) and post-translational regulation (Qin et al. 2008; Withers and Dong 2017; Augustine and Vierstra 2018; Benlloch and Maria Lois 2018). Pre-mRNA splicing is crucial for adequate plant stress responses, and while splicing is greatly influenced by different stresses (Staiger and Brown 2013; Laloum et al. 2018; Rigo et al. 2019), little is known about how stress signaling affects the splicing machinery to mediate such changes in splicing.

During pre-mRNA splicing, removal of introns is accomplished by the spliceosome, a high molecular weight complex consisting of five small nuclear ribonucleoprotein particles (snRNPs) and over 200 additional proteins (Wahl et al. 2009). The snRNPs aid in recognition of the 5′ and 3′ splice sites and contribute to the catalytic mechanism of the spliceosome. The five snRNPs contain the name-sake small nuclear uridine-rich RNAs (U1, U2, U4, U5, and U6 snRNAs). The core proteins of the U1, U2, U4, and U5 snRNPs are the Sm proteins, firstly found with antibodies from a serum lupus erythematosus patient (Stephanie Smith (Sm)). In contrast, the U6 snRNP contains the related LSM (Like Sm) proteins 2 to 8. All eukaryotes have seven highly conserved Sm proteins (B/B′, D1, D2, D3, E, F, and G) and eight LSM proteins that typically exist as hexameric or heptameric complexes in vivo. In mammalian cells, Arginine (R) methylation of Sm B/B′, D1, and D3 proteins is essential for snRNP assembly (Wahl et al. 2009). Still, in Drosophila, methylation of Sm proteins is not critical for snRNP biogenesis, highlighting species-specific differences in this basal cellular process (Gonsalvez et al. 2008).

Sm and LSM R methylation is catalyzed by protein arginine methyltransferases (PRMTs) which transfer the methyl group from S-adenosylmethionine to the guanidinium nitrogen atoms of the R residue. PRMTs are conserved from plants to humans and the importance of R methylation is underscored by impaired PRMT activity associated with autoimmune diseases or cancer in mammals. PRMTs are divided into three types, based on the kind of methylation they promote. Type I, which comprises PRMT1-4, 6, and 8, promotes monomethylation and asymmetric dimethylation of arginines (MMA and ADMA, respectively). Type II, which includes PRMT5 and 9, promotes MMA and symmetric dimethylation of arginines (sDMA). And last, type III, which consists of only PRMT7 promotes MMA of arginines (Bedford 2007).

In Arabidopsis thaliana, the best characterized PRMT is the type II methyltransferase PRMT5. Mutants of PRMT5 are late flowering, have altered circadian rhythms (Pei et al. 2007; Wang et al. 2007; Hong et al. 2010; Sanchez et al. 2010), are more susceptible to various abiotic stresses (Zhang et al. 2011; Hernando et al. 2015; Hu et al. 2017) and have widespread defects in the splicing of thousands of genes (Deng et al. 2010; Sanchez et al. 2010; Hernando et al. 2015). Whether this is due to its role as a methyltransferase remains to be elucidated.

Over the years, PRMT5 targets have been identified in plants, including the U snRNP proteins SmD1, SmD3, and LSM4, and the RNA-binding protein GLYCINE RICH PROTEIN 7 (AtGRP7), all involved in splicing (Deng et al. 2010; Streitner et al. 2012; Hu et al. 2019; Wang et al. 2020; Cao et al. 2022) suggesting that the R methylation of splicing factors could provide an efficient means to fine-tune their activity. This raised the idea that PRMT5 may affect alternative splicing (AS) of some pre-mRNAs indirectly by modulating the activity of spliceosome components through R methylation. But so far, not much is known about the direct role of R methylation in AS. The lsm4 mutant plants show severe growth retardation, are sterile and their response to salt stress and abscisic acid (ABA) is found to be severely impaired (Zhang et al. 2011). Moreover, LSM4 affects circadian clock function, likely by regulating the AS of an unknown set of core clock genes, contributing to adjustment to periodic changes in environmental conditions (Perez-Santangelo et al. 2014). Recently, LSM4 has also been implicated in the regulation of plant immunity to Pseudomonas syringae infection by interacting with the ARABIDOPSIS THALIANA METACASPASE 1 (AtMC1), modulating AS of genes such as 4-COUMARATE:COA LIGASE 3 (4CL3), a known negative regulator of plant defense (Wang et al. 2021).

In this study, we used LSM4 as a paradigm to investigate the effect of methylation of R by PRMT5 on AS. Through RNA immunoprecipitation, we identified transcripts bound by LSM4 in vivo and we show that for a suite of these targets, LSM4 regulates AS by directly binding to its transcripts. Furthermore, mRNA levels of direct targets were also affected by LSM4 binding, showing the direct role of LSM4 in both AS and steady-state mRNA levels. In addition, the lsm4 and prmt5 mutants showed a significantly high overlap of genes with altered splicing patterns, suggesting that these might be regulated by PRMT5-dependent methylation of LSM4. Using complementation studies, we show that a wild-type (WT) version, as well as an unmethylable version of LSM4, were able to rescue the lsm4-1 mutant under controlled growth conditions, suggesting that methylation is not critical for plants growing in normal growth conditions. However, LSM4 methylation changes along with stress-related environmental cues. LSM4 methylation increases upon ABA treatments and is necessary for plants to respond to salt stress. In contrast, LSM4 methylation is reduced by bacterial infection, and plants expressing an unmethylable LSM4 show a better immune response than plants expressing the WT version of LSM4. Interestingly, this phenotype is associated with our findings that, upon infection, intron retention (IR) of stress-related genes increases, diminishing the functional isoform. These data are consistent with the opposite performance of prmt5-5 and lsm4-1 mutants under biotic and abiotic stress.

Our results uncover a post-translational modification, R methylation, that can regulate a key post-transcriptional process, such as mRNA splicing, thus adding an additional layer of regulation to fine-tune antagonistic plant responses to bacterial infection and salt stress.

Results

Identification of in vivo target transcripts of LSM4

PRMT5 controls a wide range of AS events in plants (Deng et al. 2010; Hong et al. 2010; Sanchez et al. 2010; Hernando et al. 2015) and direct targets of PRMT5 methylation are enriched in RNA-binding proteins including splicing related factors such as LSM4 and AtGRP7 (Deng et al. 2010; Streitner et al. 2012; Hu et al. 2019; Wang et al. 2020). In particular, LSM4 is part of two heteroheptameric complexes, of which the LSM1 to 7 complex mainly regulates transcript stability in the cytoplasm, and the LSM2 to 8 nuclear spliceosome complex is involved in pre-mRNA splicing through U6 snRNP (Perea-Resa et al. 2012; Golisz et al. 2013). But the functional relevance of R methylation on the role of LSM4 in both the processes has not yet been described. As a first step to comprehensively understand the direct effects of LSM4 on mRNAs, we determined the transcripts bound by LSM4 in vivo. RNA immunoprecipitation followed by high-throughput sequencing (RIP-seq) was performed on plants expressing LSM4 tagged with YELLOW FLUORESCENT PROTEIN (YFP) under the control of the constitutive 35S promoter (35S:YFP-LSM4) in the lsm4-1 loss-of-function mutant. To ensure that the YFP tag does not affect LSM4 function, we confirmed the complementation of the lsm4-1 developmental phenotype (Fig. 1A). Twelve-day-old plants were crosslinked with formaldehyde to preserve in vivo RNA–protein interactions. The YFP-LSM4 fusion protein was precipitated with GFP Trap beads and co-precipitated RNAs were sequenced. We found 982 genes that were enriched in 35S:YFP-LSM4 expressing plants relative to control plants expressing 35S-YFP only (log₂ fold change ≥ |2| and false discovery rate (FDR) < 0.05), considered from now on as LSM4 targets (Fig. 1B, Supplementary Data Set 1). Of these, we found 23 genes encoding proteins directly involved in splicing, such as U1 SMALL NUCLEAR RIBONUCLEOPROTEIN-70K (U1-70K), SMD3, SKP1/ASK1-INTERACTING PROTEIN 2 (SKIP2), and SPLICING FACTOR FOR PHYTOCHROME SIGNALING (SFPS), and many other RNA-binding proteins like ALBA5 and ALBA6 (Fig. 1B). Consistently, we found an enrichment in the GO terms RNA processing, RNA splicing, and spliceosomal snRNP assembly for the direct LSM4 targets (Fig. 1C). Furthermore, in line with the reported phenotypes for lsm4-1 mutants (Zhang et al. 2011; Perez-Santangelo et al. 2014) LSM4 targets are enriched in terms related with ABA response, circadian rhythms and light responses (Fig. 1C, Supplementary Data Set 1).

Figure 1.

LSM4 and PRMT5 control an overlapping set of pre-mRNA splicing events. A) Representative photographs showing the phenotypes of plants used for RIP-seq. Plants expressing 35S:YFP-LSM4 and lsm4-1 and WT genotypes, illustrating rescue of developmental defects observed in the lsm4-1 mutant. B) Volcano plot illustrating log₂ fold-change (x-axis) and statistical significance as −log₁₀P-value (y-axis) of the RIP-seq dataset. The dashed line indicates the threshold above which transcripts are significantly enriched (FC > 3 and P-value < 0.05). C) GO-term enrichment analysis of LSM4 direct targets represented as a bubble plot. The size of the bubble is the RF. D) Relative frequencies of different AS types for all detected AS events affected in lsm4-1: Alt 3′ and Alt 5′, alternative acceptor and donor splice sites, respectively; ES, exon skipping; IR, intron retention; Multiple, for cases where more than one type of event was observed. E) Upset plot showing single, pairwise, and triple combinations of genes for the different list of genes found in each high-throughput experiment (DEG: differentially expressed gene; DU: differential usage of splicing sites from RNA-seq experiment; and targets: candidate genes bound to LSM4 found by RIP-seq). The significance of the overlap in the upset graphs was determined by calculation of the hypergeometric probability. **P-value < 0.001. F) IGV view of mapped reads for selected LSM4-bound target transcripts in 35S-GFP and 35S:YFP-LSM4 for the RIP-seq experiment and its expression in WT and lsm4-1 from RNA-seq. The gray line denotes the splicing defect as quantified by RNA-seq analysis. G, H) Venn diagram showing the extent of overlap for pre-mRNA splicing events (G) or genes with splicing events (H) affected in the lsm4-1 mutant and those altered in prmt5-5. The significance of the overlap was determined by calculation of the hypergeometric probability. I) Upset plot of several single, pairwise and triple comparisons of gene lists from diverse analyses. Statistical testing of multiset intersections was calculated using the R package SuperExactTest. ***P-value < 0.0001.

Impact of LSM4 binding on mRNA metabolism

To address the impact of LSM4 binding on its targets RNAs, we next evaluated if LSM4 targets transcripts were either differentially expressed or differentially spliced in lsm4-1 mutants. We performed RNA-seq in three biological replicates from lsm4-1 seedlings grown for 12 d under continuous light to avoid differences caused by changes in the phase of expression due to the circadian clock defect of the mutant (Perez-Santangelo et al. 2014). After mapping the reads to the Arabidopsis genome, we evaluated the splicing patterns by determining the percentage of inclusion of exons (percentage splice-in, PSI) or introns (percentage intron retention, PIR) as previously described in Mancini et al. (2021). When comparing splicing events that differed between lsm4-1 and WT plants (DU, differential usage), we obtained 6,480 altered splicing events. These corresponded to 4,499 genes (Supplementary Data Set 2), meaning that for a subset of genes, multiple splicing events were affected. We then calculated the relative distribution of the different types of AS events: alternative usage of 5′ and 3′ splice sites (alt 5′; alt 3′); IR; exon skipping (ES) and “multiple”, when more than one type was present. As reported for other splicing related proteins (Huertas et al 2019, Mateos et al. 2023, among others) IR events, the most common AS event in Arabidopsis, were the most affected (Fig. 1D). We next evaluated the frequencies of nucleotides around the 5′ splice sites (ss) and the 3′ ss for all IR events affected in lsm4-1 compared to the consensus 5′ ss or 3′ ss of all introns in the genome. We found no variation at the 5′ ss or 3′ ss sequences for introns controlled by LSM4, suggesting that LSM4 does not have a prevalent role in splice site recognition (Supplementary Fig. S1A).

When analyzing differentially expressed genes (DEGs) in lsm4-1 compare to those in WT plants, we found 3,764 genes (log₂ fold change > |1|, FDR < 0.05) (Supplementary Data Set 3). Twenty-five percent (238/982) of LSM4 targets had changed splicing patterns, while 15% (121/982) were differentially expressed (Fig. 1E), showing in plants a role of LSM4 in both processes mediated by the two LSM complexes, with a larger quantitative effect on splicing. For 33 target genes, binding of LSM4 affects both differential expression and AS (Supplementary Fig. S2A). For example, binding of LSM4 to the AT1G17990 gene leads to stabilization of its transcript as loss of LSM4 causes downregulation of the gene (Fig. 1F). At the same time, we observed increased IR of intron 4 in lsm4-1 (Fig. 1F). On the contrary, LSM4 binds to the transcript encoding the MYB59 transcription factor (AT5G59780) which likely destabilized it, as we observed increased levels of all the transcripts isoforms in lsm4-1 and LSM4 binding causes increased retention of intron 1 (Supplementary Fig. S2B).

PRMT5-mediated LSM4 methylation associates with a variety of AS events

To address a possible role of R methylation of LSM4 in splicing modulation, we compared splicing alterations in the prmt5-5 mutant obtained from our recent studies (Mateos et al. 2023) (Supplementary Data Set 4) and lsm4-1 mutants analyzed here. We reasoned that aberrant AS that occur in lsm4-1 and prmt5-5, could be due to PRMT5-mediated methylation of LSM4. Prmt5-5 showed alteration in 3,406 splicing events corresponding to 2,601 genes (Mateos et al. 2023), with 10% (730/6480) of the events affected also in lsm4-1 (Fig. 1G). Interestingly, when we compared the genes in which splicing was affected in both genotypes, the overlap increased to 30% (1348/4,499 genes) (Fig. 1H). Moreover, differently to what occurred when looking at all AS events affected in lsm4-1, the analysis of the 5′ ss of the 730 shared events disclosed a decrease in the frequency of the consensus motif for the dominant G at the −1 position and a tendency toward randomization of the nucleotides present at the −2 position (Supplementary Fig. S1B), similar to our previous results with the prmt5-5 mutant (Sanchez et al. 2010). It was previously suggested that PRMT5 modulates AS by contributing to the recognition of weak 5′ ss. Interestingly, 50% of LSM4 bound transcripts that have a splicing defect in prmt5-5 are also affected in lsm4-1 (Fig. 1I), a much higher number than the expected overlap size. Moreover, the AS affected by lsm8, a mutant of an exclusive component of the spliceosomal LSM2-8 complex (Carrasco-López et al. 2017), and lsm4 mutation (this study) cover 65% of the DU genes found in prmt5-5 (Fig. 1I). Coherently 38% of LSM4 RIP target genes show altered splicing in lsm4-1 or lsm8-1 or prmt5-5, confirming a strong molecular connection between PRMT5 and the U6 snRNP LSM2-8 complex.

Given the significant overlap in splicing changes between lsm4-1 and prmt5-5, PRMT5 could at least in part affect the function of the U6 snRNP LSM2-8 complex through methylation of LSM4. Noteworthy, our results demonstrate that LSM4 regulates AS by direct binding and we next explore if this is in part regulated by PRMT5 action.

Arginine-glycine-rich domain mutants of LSM4 rescue the lsm4-1 phenotype

To reveal to what extent the methylation of Rs in LSM4 is relevant at a molecular and physiological level, we generated transgenic plants carrying a WT version of LSM4 or an unmethylable version of LSM4 in the lsm4-1 background. LSM4 has three arginine-glycine-rich (RGG) motifs containing nine Rs in total at its C-terminus (Fig. 2A) which can be methylated by PRMT5 in vitro (Deng et al. 2010; Zhang et al. 2011). We transformed heterozygous lsm4-1 mutant plants with WT LSM4 (35S:LSM4^R) and with a version of LSM4 where the nine Rs from the three RGG motifs were changed to lysine (K) (35S:LSM4^RxK), which maintains the overall charge of the protein but cannot be methylated by PRMT5. Notably, both versions of LSM4, LSM4^R, and LSM4^RxK fully rescued the lsm4-1 phenotype, generating adult plants that resemble the WT plants in normal growth conditions (Fig. 2B). By quantifying the leaf size area of seedlings grown on soil for 12, 15 and 20 d, we observed that the transgenic lines were slightly larger than WT plants, although the differences were not significant. This indicates that methylation status does not affect plant development during normal growth (Fig. 2C). To test if changes from Rs to Ks indeed prevent symmetric dimethylation of the LSM4 protein, we performed western blots using protein extracts from LSM4^R and LSM^RxK seedlings probed with the SYM10 antibody directed against symmetric dimethyl-arginine. Extracts from prmt5-5 plants that have globally impaired symmetric dimethylation of Rs served as negative control (Mateos et al. 2023). We observed a signal at ∼15 kDa in extracts from plants expressing the WT LSM4^R protein which was absent in prmt5-5 extracts (Fig. 2D). Similar to prmt5-5, LSM4^RxK plants showed no methylation signal (Fig. 2D), suggesting that mutating Rs to Ks bypasses methylation of LSM4. As no anti-LSM4 antibody is available to verify that the band indeed corresponds to LSM4 and to confirm that the absence of signal in LSM4^RxK is not due to altered protein abundance, we generated plants expressing variants of YFP-LSM4. As in plants used for our RIP-seq experiment, we inserted the LSM4^RxK version downstream of the YFP tag. Western blots confirmed the results we observed with the untagged lines, where methylation is prevented by changing Rs to Ks (Fig. 2E). The expression level of both versions of LSM4 is similar as detected with the GFP antibody (Fig. 2E). We previously observed that the lsm4-1 mutant had a long-period phenotype (Perez-Santangelo et al. 2014). Given the strong developmental defects in lsm4-1, leaf movements experiments are not viable. Nevertheless, the circadian period of expression of clock genes, such as the CCA1 and CCR2 genes, was lengthened by more than 2 h in lsm4-1 homozygous mutants relative to WT plants (Perez-Santangelo et al. 2014). This highlights the role of LSM4 in maintaining the correct function of the clock. We next analyzed if methylation of LSM4 is needed to ensure proper clock function. As the LSM4^R and LSM4^RxK plants in the lsm4-1 background showed proper development at a seedling stage, this time we quantified the effect on the clock by measurement of rhythms in leaf movement. We analyzed both versions of LSM4, LSM4^R, and LSM4^RxK in the lsm4-1 background compared to in the WT and found that neither of them showed the distinct long-period phenotype as expected for lsm4-1 (Fig. 2F), suggesting that the methylation status of LSM4 does not impact the role of LSM4 on circadian rhythm. Taken together, these results demonstrate that the Rs of all three RGG domains of LSM4 are methylated in vivo and that changing these residues to Ks prevents methylation without affecting protein stability. More importantly, LSM4 methylation is not critical for rescuing the phenotype under normal growth conditions.

Figure 2.

The RGG domains of LSM4 are not required for lsm4-1 phenotype rescue. A) Scheme showing wild-type (WT) (LSM4^R) and mutant (LSM4^RxK) LSM4 proteins. RGG domains are shown as boxes. B) Representative pictures of WT, lsm4-1, and two independent lines for each type of transgenic plant, LSM4^R and LSM4^RxK. Twelve-days-old seedlings (top panel) and 4-wk-old plants (bottom panel). C) Leaf size of seedlings from 12, 15, and 20 d grown in long-days at 22 °C. At least 25 plants were measured per genotype. Mean and Sd are plotted. D) Methylation of LSM4 protein is impaired by R to K changes. Immunoblot of LSM4^R, LSM4^RxK, and prmt5-5 samples with the SYM10 antibody. The prmt5-5 mutant was used as a negative control for impaired methylation. Ponceau red staining shows equal loading across samples. The arrow represents the expected size for LSM4. E) Immunoblot of WT, YFP-LSM4^R, and YFP-LSM4^RxK samples developed with SYM10 (top) or anti-GFP antibody (bottom). The arrow represents the expected size for the YFP-LSM4 fusion protein. Ponceau red staining shows equal loading across samples. F) Circadian rhythm of leaf movement. Plants’ vertical leaf motion (RLM) was obtained for the first pair of leaves of seedlings entrained under long-day conditions (16 h light/8 h dark) and then transferred to continuous light (LL). WT, wild-type, LSM4^R, three independent lines from lsm4-1 mutant plants transformed with LSM4^R and LSM4^RxK, three independent lines from lsm4-1 mutant plants transformed with LSM4^RxK. (Left) Period length of leaf movement rhythms is estimated by Fast Fourier transform–nonlinear least-square test (FFT–NLLS) (Right). Error bars represent SEM.

LSM4 methylation impacts AS of genes associated with stress response

The large overlap of genes with splicing changes between prmt5-5 and lsm4-1 mutants (RF 3.9, P < 6.915e−233) suggests that the activity of LSM4 in controlling a subset of splicing events might be mediated in part by its methylation in the RGG domains (Fig. 1, H and I). To directly address this, we performed RNA-seq experiments of the transgenic lines expressing the methylable LSM4^R or the unmethylable LSM4^RxK in the lsm4-1 background. To analyze our data, we performed three different pairwise comparisons: LSM4^R in lsm4-1 vs lsm4-1; LSM4^RxK in lsm4-1 vs lsm4-1 and lsm4-1 vs WT. More than 90% of the DEGs in lsm4-1 overlapped with those of either LSM4^R or LSM4^RxK lines, highlighting that the transcriptomes of LSM4^R and LSM4^RxK are much alike (Fig. 3A, Supplementary Data Sets 3, 5, and 6), which is also associated with the ability of both versions of LSM4 to rescue the lsm4-1 phenotype under normal growth conditions. Most importantly, when we focus on the impact of the LSM4 methylation status on AS, we found that of the 6,480 events altered in lsm4-1 relative to in WT (Supplementary Data Set 2), 4,888 were shared with either LSM4^R or LSM4^RxK when comparing with lsm4-1, reaching 75% of the total events (Fig. 3A, Supplementary Data Sets 7, 8) showing that methylation of LSM4 has a larger effect on splicing than on transcript levels. As for expression, we observed that both LSM4 versions, LSM4^R and LSM4^RxK, largely complement the effect of the mutation in splicing resembling the pattern in WT, but still they show slight variations among them, suggesting that methylation status might have a role in AS (Supplementary Fig. S3).

Figure 3.

LSM4 R methylation affects a subset of transcripts involved in stress responses. A) Summary of the transcriptome analysis is shown as a Circos plot. The total number of DEGs (left) or affected splicing events (right) in each mutant is shown below the mutant name. Numbers of upregulated or downregulated genes are indicated with arrows. Connecting lines are scaled and represent shared affected events. Blue lines shared across all genotypes; green lines shared between LSM4^RxK and lsm4-1; light violet, shared between LSM4^RxK and LSM4^R, and dark violet shared between LSM4^R and lsm4-1. Numbers above the lines represent the number of genes (left) or events (right). B) Scheme showing the rationale behind analysis to define splicing events dependent on LSM4 methylation. C) The effect of lsm4-1 mutation is larger than the effect of its methylation. ΔPSI/PIR values for the splicing events affected by LSM4 methylation in lsm4-1 vs WT or LSM4^R vs LSM^RxK. The number of events is shown in parentheses. D) IGV view of mapped reads for selected transcripts affected by R methylation in WT, lsm4-1, LSM4^R, and LSM4^RxK. The vertical line denotes the differential splicing event as quantified by RNA-seq analysis. The PIR value is shown for each genotype. E) GO-term enrichment analysis of genes affected by R methylation represented as a bubble plot. The size of the bubble is the RF.

To examine more precisely the role of LSM4 methylation in splicing, we defined events that depend on LSM4 methylation. For this, we calculated the PSI and PIR values for the LSM4^R and LSM4^RxK genotypes of the 6,480 splicing events affected in lsm4-1 (Fig. 3B). Next, we selected events that differ in their PSI or PIR between LSM4^R and LSM4^RxK by more than 3% (Xin et al. 2017). We found 928 splicing events, corresponding to 846 genes, that we called events controlled by LSM4 dependent on R methylation (Fig. 3B, Supplementary Data Set 9). Of these, we first focused on the 827 IR events, where we observed a significant change in the frequency of the G nucleotide at position −1 at 5′ ss, although to a lesser extent than observed for prmt5-5 (Supplementary Fig. S1C). Interestingly, when we compared the effect of the lsm4-1 mutation with that of LSM4 methylation, we found three important characteristics. Firstly, the effect of mutating LSM4 on splicing is larger than that of its methylation, as judged by the higher PSI or PIR values calculated for the same set of shared splicing events (Fig. 3C). Secondly, the lsm4-1 mutation is always detrimental to splicing, meaning that there are more retained introns in the mutant than in the WT. Lastly, strikingly, LSM4 methylation has a differential effect on IR events. We found that ∼70% of the IR events in lsm4-1 restored to a larger extent the effect of the lsm4 mutation in LSM4^RxK than in LSM4^R (Fig. 3C), as shown for AT1G35350 (Fig. 3D) suggesting that methylation could be inhibitory for certain splicing events. On the contrary, in the case of AT5G37500 transcript, R methylation has a positive effect on splicing (Fig. 3D) as LSM4^R fully complements IR defects present in lsm4-1 while LSM4^RxK plants fail to restore proper splicing.

To begin to understand the biological processes that are affected by LSM4 methylation, we focused on AS events with PSI or PIR difference between LSM4^R and LSM4^RxK (PSI or PIR difference between genotypes >5%, Supplementary Data Set 9, highlighted in blue). We tested this subset of events for overrepresented functional categories by GO term enrichment analysis. Intriguingly, the most significant functional categories were associated with immune responses, including defense-associated salicylic and jasmonic acid hormonal signaling (Fig. 3E). In addition, response to drought, salinity, and ABA signaling appear as enriched categories, suggesting that LSM4 methylation has a minor effect on plant growth and development, but might be relevant when plants are exposed to adverse conditions (Fig. 3E). When we classified the genes according to the effect on splicing, we observed that the genes from LSM4^RxK whose events were more similar to WT than the LSM4^R to WT were almost exclusively enriched in defense-related categories. Collectively, we found that LSM4 methylation plays an important role in modulating gene expression and splicing patterns of a subset of genes associated with biotic and abiotic stress responses, showing a possible role of LSM4 R methylation in the response to these stresses.

Impaired methylation of LSM4 leads to hypersensitivity to ABA and salt stress

Given that the terms “response to abscisic acid” and “response to abiotic stress” were enriched in our GO terms analysis of genes with splicing events dependent on LSM4 methylation (Fig. 3E), we set out to study the effect of LSM4 methylation during abiotic stress. Considering that the increase of ABA levels is a primary signal induced during the adaptive response to abiotic stresses, including drought and salinity (Yang et al. 2017) we studied the response of seedlings of the transgenic lines LSM4^R and LSM4^RxK under ABA, sorbitol, and NaCl treatments. We observed a complex pattern of response to abiotic stress. We saw that at very early stages (2 d), plants overexpressing the WT version of LSM4 germinate less efficiently than WT or LSM4^RxK plants either during ABA or sorbitol treatments (Fig. 4, A and E). Immediately after germination, ABA is associated with a developmental checkpoint inducing growth arrest when young seedlings are stressed (Lopez-Molina et al. 2001). Once germinated, the methylation status of LSM4 has a positive effect upon abiotic stress. A reduced percentage of seedlings of LSM4^RxK lines reach the fully open cotyledon stage (known as greening) under ABA or sorbitol treatment compared to LSM4^R lines (Fig. 4, B, C, and E, Supplementary Fig. S4A). As it was previously shown (Zhang et al 2011), prmt5-5 is more sensitive to ABA (Fig. 4B). Consistent with this, unmethylated LSM4 shows enhanced sensitivity to ABA as prmt5-5 (Fig. 4B). In addition, LSM4^RxK seedlings displayed a strong decrease in fresh weight (FW) after 7 d on 1 µM ABA compared to LSM4^R and the WT (Fig. 4D). Similarly, LSM4^RxK seedlings when grown on MS plates supplemented with 200 mm sorbitol showed reduced growth and FW, but enhanced accumulation of anthocyanin (Fig. 4, F and G, Supplementary Fig. S4B). LSM4^RxK seedlings exposed to 50 mm NaCl also showed a reduction in FW and a reduced number of lateral roots (LRs) compared to LSM4^R (Fig. 4, H and I), highlighting that the lack of methylation is detrimental for abiotic stress responses. No differences between lines were detected at primary root (PR) elongation (Supplementary Fig. S4C). In addition to growth retardation, when plants are exposed to salinity, they suffer chlorosis. When treated with 75 mm NaCl, LSM4^RxK seedlings were more sensitive to salt stress in terms of loss of chlorophyll content compared to LSM4^R (Fig. 4J).

Figure 4.

LSM4 R methylation is required for abiotic stress responses. Time-course of germination (A) and greening (B) percentage for wild type (WT), prmt5-5, transgenic lines of 35S-LSM4^R, and transgenic lines of 35S-LSM4^RxK. Seeds were sown on MS plates (control) or MS plates supplemented with 1 µm ABA. Percentages are relative to control for each genotype at each point. C) Representative seedlings grown on 1 µm ABA for 7 d. D) Four-days-old seedlings were transferred to new MS plates (as control) or to MS supplemented with 1 µm ABA. FW was quantified 7 d post-treatment, and expressed relative to the control for each genotype. E) Percentage of germination and greening after 2 or 8 d in 250 mm sorbitol, respectively. Percentages are relative to control for each genotype at each time-point. F) Representative seedlings grown on sorbitol treatment for 14 d. G) Anthocyanin accumulation of control or 250 mm sorbitol treated seedlings. H and I) Four-days-old seedlings were transferred to 50 mm NaCl and FW (H) and LR) (I) were analyzed 7 d post-treatment. J) Chlorophyll content of plants treated with 75 mm NaCl for 7 d. NaCl data is expressed relative to the control for each genotype. For all experiments depicted in A to J). Data are means ± Sd of nine biological replicates. Different letters indicate significant differences among genotypes, P < 0.05 (one-way ANOVA followed by Tukey's multi-comparison test). K) Methylation of LSM4 protein is changed upon ABA treatment. Immunoblot of LSM4^R samples after 0, 1, or 4 h of 10 µm ABA treatment developed with the SYM10 antibody. The LSM4^RxK mutant was used as a negative control for impaired methylation. Ponceau red staining shows equal loading across samples. The arrow represents the expected size for LSM4.). Data are means ± Sd of three independent experiments. L) IGV view of mapped reads for selected transcripts involved in ABA responses affected by R methylation in WT, lsm4-1, LSM4^R, and LSM4^RxK. The vertical line denotes the splicing defect as quantified by RNA-Seq analysis. The PIR value is shown for each genotype.

To further explore the dynamics of LSM4 methylation in vivo, we used the SYM10 antibody to test sDMA levels in protein extracts from LSM4^R seedlings incubated in Murashige and Skoog (MS) medium supplemented with ABA. An induction of LSM4 methylation was detected 4 h post-treatment in LSM4^R seedlings in vivo (Fig. 4K). LSM4^RxK was used as a negative control. Abiotic treatments had no effect on protein levels, either on methylated or unmethylated LSM4 (Supplementary Fig. S4D). Previous studies reported that the R methylation signal of some proteins of ∼14 kDa was increased in WT seedlings subjected to salt and ABA treatments (Zhang et al. 2011; Hu et al. 2017) and that nitrosogluthathione enhanced PRMT5 methyltransferase activity in vitro using histone4 and LSM4 as substrates (Hu et al. 2017). Here, our results show that the hypersensitivity to ABA of seedlings impaired in LSM4 methylation correlates with the increase in LSM4 methylation, suggesting that methylation of this U6 snRNP component could play a role in plant responses to abiotic stress mediated by ABA signaling.

LSM4 methylation differentially controls AS of specific introns from selected abiotic stress-related pre-mRNAs

Since the LSM4^RxK lines displayed a different response to ABA and salinity than the methylated LSM4^R and WT seedlings, we next explored splicing patterns of genes related to both treatments to determine a possible molecular mechanism behind. Genes selected from GO categories of drought, salinity or ABA response, such as ABA-HYPERSENSITIVE GERMINATION 2 (AHG2) and cell outward potassium channel (GORK) showed IR events that are differentially affected in LSM4^R compared to LSM4^RxK lines, indicating that these IRs are dependent on the methylation status of LSM4 (Fig. 4L, Supplementary Fig. S5). AHG2 is induced by ABA and the corresponding mutant is hypersensitive to ABA in germination assays (Lopez-Molina et al. 2001; Hirayama et al. 2013), while the mutant for GORK gene has an increased water consumption and reduced FW in response to ABA, similar to the phenotypes reported here for LSM4^RxK seedlings (Fig. 4, A to J). Therefore, we propose that impaired splicing of these genes in LSM4^RxK lines may contribute to ABA hypersensitivity. Furthermore, we analyzed other genes involved in ABA signaling and related to abiotic stress responses, finding that RESPONSIVE TO DESICCATION 22 (RD22) splicing is regulated by PRMT5 but not by LSM4. While splicing of ABSCISIC ACID RESPONSIVE ELEMENTS-BINDING FACTOR 2 (ABF2; AT1G45249) and HYPERSENSITIVE TO ABA1 (HAB1; AT1G72770) are regulated by LSM4 (Supplementary Fig. S6) similar to what was found by Zhang et al. (2011), but only ABF2 seems to be dependent on LSM4 methylation. Taken together, LSM4 methylation regulates intron splicing of genes associated with ABA, such as AHG2, and ABF2, which may contribute to plant adaptation to abiotic stress including salinity and drought.

Methylation of LSM4 negatively regulates splicing of defence genes and plant immunity

Arabidopsis plants respond to bacterial infection via two connected branches of immune signaling. The pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) is initiated when plasma membrane-localized pattern recognition receptors detect the corresponding PAMPs; while effector-triggered immunity (ETI) is a stronger response and is activated through recognition of virulence factors by intracellular immune receptors including the nucleotide-binding leucine-rich repeat (TIR-NB-LRR) protein family, and induces specific and localized immune reactions (Boller and Felix 2009; Nakai et al. 2013). Since our GO analysis of genes with methylation-dependent splicing events of LSM4 showed an enrichment of the terms “immune responses”, including defense-associated “salicylic and jasmonic acid hormonal signaling” (Fig. 3E), we focused on whether LSM4 methylation has a regulatory role during bacterial infection. We first compared genes differentially spliced dependent upon the LSM4 methylation status (Supplementary Data Set 9) with genes differentially spliced in WT plants after infection with P. syringae pv. tomato DC3000 (Pst) (Golisz et al. 2021). For 113 genes splicing is affected by the LSM4 methylation status as well as Pst infection (Fig. 5A, Supplementary Data Set 10), implying that R methylation can modulate bacterial resistance by regulating the splicing of specific transcripts. Interestingly, 87 out of 113 genes had increased levels of the IR isoform in the LSM4^R genotype judged by their PIR value (Supplementary Data Set 10, Fig. 5B). LSM4^R had increased IR of VASCULAR PLANT ONE ZINC FINGER PROTEIN (VOZ1) and RADICAL INDUCED CELL DEATH1 (RCD1) mRNAs, both required for the activation of the immune responses to bacterial and fungal pathogens in Arabidopsis (Bardou et al. 2014; D’Ippolito et al. 2016; Wirthmueller et al. 2018). In contrast, IR of the IAA-LEUCINE-RESISTANT (ILR1)-LIKE 3 (ILL3) member of the amidohydrolase family responsible for increasing the levels of free active IAA and negatively associated with defense (D’Ippolito et al. 2016) is higher in LSM4^RxK seedlings compared to IR in LSM4^R seedlings (Supplementary Data Set 10). So far, these results indicate that LSM4 methylation modulates IR events of genes involved in plant immunity by regulating the ratio of isoforms to increase or decrease the functional isoform depending on its activity in plant defense. In agreement, a significant number of the introns differentially targeted by LSM4 methylation were in transcripts encoding defense signaling proteins associated with the ETI branch of the immune response (Fig. 3E). In all cases, IR in these defense signaling genes was higher in LSM4^R than in unmethylable LSM4^RxK plants. For instance, five members of the TIR-NBS-LRR family show increased IR in LSM4^R compared to WT, while LSM4^RxK plants show more efficient splicing almost indistinguishable from WT plants (Fig. 5C). For all of these genes, IR causes a premature stop codon leading to shorter proteins (Supplementary Fig. S7), meaning that in LSM4^R plants, intron-containing transcripts would encode premature stop codons that may result in truncated proteins with potentially diminished functions. Furthermore, ZYGOTIC ARREST 1 (ZAR1), a key NLR protein that is a calcium-permeable cation channel for triggering immunity and cell death, shows increased IR in plants expressing LSM4^R. Interestingly, unmethylable LSM4^RxK plants show perfect splicing of intron 1, increasing the level of the functional isoform that upregulates plant defenses (Fig. 5D). These results suggest that LSM4 methylation could be a means for controlling proper splicing of defense genes during bacterial infection. To test this, leaves from 2-wk-old LSM4^R plants were infiltrated with the bacterial pathogen Pst or MgSO₄ as mock. We observed that 24 h post-infection (hpi), plants infected with Pst displayed a reduction in LSM4 methylation compared to mock-treated plants as evidenced by immunoblots with SYM10 antibody (Fig. 5E). An early increment of LSM4 methylation 4 hpi, which is opposite to the 24 h response, was also detected. Still, this induction could be triggered by bacteria and counteracted by the plant to overcome the infection. This temporal pattern was reported for the accumulation of the phytohormone auxin which is induced by most pathovars of Pst (Thilmony et al. 2006; Chen et al. 2007), while auxin signaling is repressed by Arabidopsis plants to reduce bacterial infection (Navarro et al. 2006).

Figure 5.

LSM4 methylation modulates plant immunity. A) Venn diagram showing the extent of overlap for pre-mRNA splicing events affected by LSM4 methylation and those altered upon bacterial infection from Golisz et al. (2021). B) Heatmap showing ΔPSI/PIR values calculated for the 113 genes from (A) in LSM4^RxK vs LSM4^R. C) IGV view of mapped reads for selected transcripts coding for TIR-NBS-LRR proteins in WT, lsm4-1, LSM4^R, and LSM4^RxK. The vertical line denotes the splicing defect as quantified by RNA-seq analysis. The PIR value is shown for each genotype. D) IGV view of mapped reads for ZAR1 in WT, lsm4-1, LSM4^R, and LSM4^RxK. The vertical line denotes the splicing defect as quantified by RNA-seq analysis. E) (Left) Methylation of LSM4 protein upon mock treatment (M) or 4 and 24 h of bacterial infection (I) with P. syringae DC3000. Immunoblot of LSM4^R samples with the SYM10 antibody. The LSM4^RxK mutant was used as a negative control for impaired methylation. Ponceau red staining shows equal loading across samples. Arrow represents expected size for LSM4. (Right) Quantitation of three independent experiments. Error bars represent Sd (**P < 0.001; t-test to mock). F, G) RT-PCR to detect splicing defects for AT1G56520 and AT1G72900in LSM4^R and LSM4^RxK upon bacterial infection (I) or mock treatment (M). The WT and lsm4-1 genotypes were measured under controlled conditions to show IR is increased in lsm4-1. Alternative regions are highlighted in red in the diagrams next to the gels and position of the primers used are depicted. The ratio of the splicing is shown below is lane. H) Four-weeks-old plants grown in LD conditions were infected by infiltration with P. syringae DC3000. Bacterial growth was assessed 2 dpi. (CFU: colony-forming units) in WT, two independent lines of LSM4^R, two independent lines of LSM4^RxK plants. Data represent the average of log-transformed bacterial growth (n = 8 independent biological replicates). This experiment was repeated twice with similar results. Error bars indicate SEM. P < 0.05 (one-way ANOVA followed by Tukey's multicomparison test). I and J) Levels of PR1 (I) and PR2 (J) proteins upon bacterial infection (I) or mock treatment (M) determined by western blots. PR1 levels were determined in LSM4^R and LSM4^RxK plants after 24 or 48 h upon bacterial infection. Levels of PR2 were measured after 24 h of infection. Anti-PR1 and anti-PR2 antibodies were used, respectively. Ponceau red staining shows equal loading across samples.

To better understand the potential molecular consequences on splicing of reduced methylation after infection, we next tested splicing patterns upon 24 h of bacterial infection in two genes of the TIR-NBS-LRR family (AT1G56520 and AT1G72900, Fig. 5C). In normal conditions, lsm4-1 mutant presents an increased IR in the AT1G56520 transcript, which can be visualized by RT-PCR with primers binding to exon 3 and 4 expanding intron 3, compared to WT plants (Fig. 5F). Upon Pst infection, IR increased in LSM4^R while LSM4^RxK showed WT levels of spliced intron (Fig. 5F). Similar results were observed when analyzing IR in intron 1 of the AT1G72900 gene, which is again increased in LSM4^R compared with LSM4^RxK under Pst infection (Fig. 5G). These differences in splicing patterns are given by the methylation status of LSM4 and not by variance in expression of the transgenes, as the LSM4 transcript accumulates to similar levels in the LSM4^R and LSM4^RxK lines analyzed here (Supplementary Fig. S8). The function of the rest of the proteins in plant immunity and the possible effect of IR on their function remain to be analyzed.

We reasoned that if increased methylation of LSM4 is detrimental to proper splicing of plant defense genes on one hand (Fig. 5, F and G) and plant responses to bacterial infection tend to minimize LSM4 methylation in part by reduced levels of PRMT5 on the other hand (Hong et al. 2010), this should translate into a differential activation of immune responses in LSM4^R and LSM4^RxK plants. To assess this, LSM4^R and LSM4^RxK plants were challenged with the pathogen Pst. Leaves were infiltrated with Pst and bacterial growth was analyzed 2 d post-infection. We found that unmethylated LSM4^RxK lines developed less susceptibility to Pst infection compared to LSM4^R or WT plants (Fig. 5H). This is antagonistic to the abiotic stress response where LSM4^R seedlings show better performance compared to unmethylated LSM4^RxK seedlings. The enhanced resistance of the LSM4^RxK plants was associated with an increased basal expression of the typical marker of the salicylic acid-mediated defense system, PATHOGENESIS-RELATED 1 (PR1) at the mRNA and protein levels (Fig. 5I, Supplementary Fig. S9). In addition, an early response of PATHOGENESIS RELATED PROTEIN 2 (PR2) was detected at 24 hpi in LSM4^RxK infected plants compared to LSM4^R (Fig. 5J). Wang et al. (2021) found that plants overexpressing LSM4 were more susceptible to bacterial infection which associated with a reduced level of PR1 in LSM4 overexpressing plants compared to in WT plants upon infection. Although these results are in line with our observations, we could not detect an increased susceptibility of LSM4^R plants to Pst infection compared to the WT, probably due to differences in growth conditions. Notably, both results indicate that higher LSM4 levels or increased methylation status seem to downregulate plant immunity and PR activation. This aligns with the effect of R methylation on ARGONAUTE2 (AGO2) by PRMT5 where plants lacking the RGG domain of AGO2 are more resistant to bacterial infection (Hu et al. 2019). Indeed, similar to LSM4^RxK plants, prmt5-5 mutants showed a weak but enhanced response to bacterial infection (Hu et al. 2019). This downregulation of PRMT5 during infection could entail reduced LSM4 methylation. Our results show that the plant response to biotic stress could be mediated by splicing regulation of defense response genes through modulation of the LSM4 methylation status.

Discussion

Alternative pre-mRNA splicing is an important mechanism to rapidly adjust the transcriptome to environmental changes, especially allowing rapid adaptation to stresses without the obligation/need of de novo transcription (Staiger and Brown 2013, among others). R methylation of spliceosomal RBPs is a well-documented post-translational modification (Deng et al. 2010; Hu et al. 2019; Cao et al. 2022). Yet, experimental evidence showing its impact on splicing was still lacking. Using LSM4, an R-methylable U6 snRNP component of the spliceosome as a paradigm for studying RBP methylation, allowed us to uncover an important regulatory layer involved in fine-tuning plant responses to abiotic and biotic stress in different ways.

Using RIP-seq, we identified 982 target transcripts of LSM4, many of which are coding for splicing-related proteins or RBPs. As RIP is biased to highly abundant transcripts, other RNAs might also be direct targets of LSM4 and could not be detected in this study. Cross-referencing the targets with the transcriptome of the lsm4-1 mutant revealed that for ∼25% of the LSM4 targets splicing was altered, whereas for ∼15% expression levels were changed. Thus, we show that LSM4 is involved in both AS and steady-state levels of transcripts by directly binding to RNAs, a distinguishing feature that was only described for a small number of RBPs in plants (Staiger et al. 2003; Day et al. 2012; Bardou et al. 2014; Wu et al. 2016). The glycine-rich domain of LSM4 was shown to enhance mRNA stability and alter mRNA decay pathway in Saccharomyces cerevisiae, but its action is not by binding to the transcripts but by association with the decapping activator Edc3 (Huch et al. 2016).

A close analysis of the LSM4 targets suggests that LSM4 regulates AS and/or mRNA levels of several other RBPs involved in post-transcriptional processes, suggesting a feedback/cooperation among RBP complexes to ensure a proper transcriptome. Of the LSM4 target transcripts that are as well differentially expressed in the mutant, a similar proportion was up- or downregulated (Supplementary Fig. S10), suggesting that LSM4 binding can lead to both stabilization or destabilization of transcripts without any preference. This does not seem to be a general rule for other RBPs involved in splicing such as AtGRP7, as in this case, binding reveals a predominantly negative effect of AtGRP7 on its targets (Meyer et al 2017).

A possible role of R methylation of LSM4 by PRMT5 in modulating splicing should be reflected by AS changes occurring in both lsm4-1 and prmt5-5. We found that around 30% of genes differentially spliced in lsm4-1 are also differentially spliced in prmt5-5. Thus, PRMT5 could at least partially affect the function of the U6 snRNP LSM2-8 complex through LSM4 methylation. Methylation of other LSM components of the complex seems less likely. For instance, we could not detect sDMAs on LSM8 (Supplementary Fig. S11). Previously, LSM4, LSM7, LSM6B, and LSM8 were detected by mass spectrometry of proteins immunoprecipitated with sDMA-specific SYM11 antibody (Boisvert et al. 2003), but methylation by PRMT5 in vitro was only confirmed for LSM4 (Zhang et al. 2011). Therefore, Zhang et al. (2011) proposed that the unmethylated LSM proteins co-immunoprecipitated nonspecifically with methylated LSM4. We cannot exclude a role of PRMT5 in splicing independent from its methylation activity, for instance, acting as a scaffold protein of the splicing complex. However, the effect of the prmt5-5 mutation was larger on LSM4 methylation-dependent than on methylation-independent genes, as judged by the PIR values (Supplementary Fig. S12A), reinforcing the idea that methylation of LSM4 by PRMT5 fine-tunes splicing. In line with this, 40% of LSM4 direct targets that depend on methylation for their splicing are also affected in prmt5-5, while PRMT5 mutation only controls 20% of the nonmethylation dependent targets (Supplementary Fig. S12B).

LSM4 methylation by PRMT5 may not be completely essential for the subset of splicing events regulated by LSM4 through the U6 snRNP complex since many genes are spliced correctly independently of PRMT5 action (Fig. 1, G and H). Rather, sDMAs on LSM4 could constitute an extra layer of control that contributes to splicing of specific transcripts, particularly those with weak 5′ ss. It should not be discarded that some of the events affected in lsm4-1 might be an indirect action caused by either altered gene expression, as LSM4 is as well part of the LSM1-7 complex controlling mRNA stability, or by variations on splicing pattern of splicing-related factors due to the lsm4-1 mutation. In particular, R methylation of LSM4 by PRMT5 does not appear to be a requirement for splicing modulation by the U6 snRNP complex under normal growth conditions. This conclusion extends to plants expressing LSM4 under its own promoter as we observed that both LSM4 versions complement the lsm4-1 mutation (Supplementary Fig. S13, A and B). Moreover, both versions of LSM4:LSM4 plants can restore splicing defects found in lsm4-1 to a similar extent as we observed for the overexpressing lines (Supplementary Fig. S13, C and D). The increased IR on AT4G26520 affected in lsm4-1 was equally restored by LSM4:LSM4^R and LSM4:LSM4^RxK, while defects on AT2G23420 were fully complemented by LSM4:LSM4^R but to a lesser extent by LSM4:LSM4^RxK (Supplementary Fig. S13C). On the contrary, we show that treatments associated with stress are capable of modulating LSM4 methylation levels precisely. LSM4 methylation is beneficial during salinity and abiotic stress (Fig. 4) while responses to bacterial infection are favored when methylation levels are limited (Fig. 5). This is consistent with the opposite performance of prmt5-5 and lsm4-1 mutants under the same stresses (Zhang et al. 2011). Our work suggests that increased R methylation is advantageous for plants under ABA, sorbitol, or NaCl treatments typically associated with abiotic stresses. Several pieces of evidence support our idea. Firstly, LSM4 sDMA is favored upon ABA treatments (Fig. 4K). Secondly, unmethylable LSM4 plants show hypersensitivity when facing abiotic stress (Fig. 4, A to J, Supplementary Fig. S4). Lastly but most importantly, we found that LSM4 methylation regulates intron splicing of several genes associated with ABA such as AHG2 and ABF2. In abiotic stress IR is increased in LSM4^RxK compared to LSM4^R or WT, indicating that improper splicing of these genes might be responsible for the hypersensitive phenotype (Fig. 4L). Furthermore, analysis of the lsm8 mutant revealed that the U6 snRNP LSM2-8 complex is necessary for the precise splicing of a set of stress-related pre-mRNAs, related to cold and salt stress tolerance (Carrasco-López et al. 2017). Moreover, mutants in PRMT5, LSM4, and LSM5 are hypersensitive to salinity and ABA due to inaccurate splicing of stress-related genes (Zhang et al. 2011; Cui et al. 2014).

Our results are also in agreement with multiple pieces of evidence associating different components of the spliceosome with redundant and complex regulation of splicing under abiotic stress. The Sm core protein SmEb is induced by ABA treatment, and also modulates the AS of HAB1 similar to the SKIP splicing factor (Huertas et al. 2019; Hong et al 2021; Zhang et al 2022). In addition, an analysis of thousands of public RNA-seq libraries selected PRMT5 and SKIP as the main regulators of introns present in nuclear-unspliced chromatin-bound polyadenylated transcripts, called post-transcriptionally spliced (pts) introns (Jia et al. 2020). Regulation of pts introns by PRMT5 and SKIP could constitute a rapid mechanism to produce fully spliced functional mRNAs for plants to rapidly adapt to changing environmental conditions where pts introns are enriched. We propose that this action of PRMT5 could be in part due to its role as methyltransferase of RBPs.

Contrarily, biotic stress negatively regulates LSM4 methylation, evidenced by reduced SYM10 signal of LSM4 upon bacterial infection (Fig. 5E). Consequently, plants expressing unmethylable LSM4 show stronger resistance to Pst (Fig. 5). A recent study also demonstrates that overexpression of LSM4 renders plants more susceptible to Pseudomonas than WT plants (Wang et al. 2021), in line with the idea that increased LSM4 methylation is detrimental for plant immunity. All this is in accordance with previous observations that PRMT5 transcript levels are slightly reduced during infection with Pst (Hu et al. 2019). In fact, R methylation of AGO2 protein by PRMT5 has been proposed to participate in defense responses. Downregulation of PRMT5 during infection causes reduced AGO2 R methylation, leading to stabilization of AGO2 and AGO2-associated sRNAs to promote plant immunity (Hu et al 2019). Along with decreased LSM4 methylation levels, we observed that splicing of defense-related genes is also influenced by the methylation status (Fig. 5). In general, we found IR of genes involved in plant immunity to be altered following the LSM4 methylation pattern. Interestingly, the IR ratio between isoforms does not always shuffle in the same direction for all genes. For instance, negative regulators of immunity have more IR, while positive regulators have less IR in unmethylable plants or after pathogen entry. But still how and whether an intron is selected for increased or diminished IR upon LSM4 methylation is a question yet to be answered.

Taken together, this study provides direct evidence that R-methylation of a component of U6 snRNP forms an additional layer in post-translational regulation, with antagonistic roles in the plant response to abiotic and biotic stress, respectively (Fig. 6). Furthermore, LSM4 methylation was associated with splicing changes. Thus, PRMT5-mediated methylation of LSM4 has an impact on several stress responses associated with the splicing control of genes involved in these stresses.

Figure 6.

Proposed model for the role of LSM4 methylation in modulation of stress responses. A) In the absence of lsm4 a wide range of AS defects leads to aberrant phenotypes and altered stress responses. Splicing defects are enriched in IR events. B) Upon exposure to bacterial infection, PRMT5 decreases (Hu et al. 2019) and methylation of LSM4 is reduced (this work). This reduction affects splicing of genes involved in plant immunity to increase bacterial resistance. C) Treatments associated with abiotic stress increase LSM4 methylation (this work and Hu et al. 2017). A decrease in IR of genes involved in abiotic stress response leads to an increase of the functional isoforms correlating with improved adaptation to abiotic stress. Created with BioRender.com (Agreement Number QR26I9QC8X).

Materials and methods

Plant material and growth conditions

All Arabidopsis (A. thaliana) mutants are in the Columbia-0 (Col-0) ecotype. The lsm4-1 (SALK_063398) and prmt5-5 mutants were previously described (Sanchez et al. 2010; Perez-Santangelo et al. 2014). Plants were grown in soil at 22 °C under long days (LD cycles (16 h light/8 h dark); 80 μmol m⁻² s⁻¹ of white light) for infection assays. For the rest of the experiments, seeds were surface sterilized, sown in MS medium containing 0.8% (w/v) agar and stratified for 3 d in darkness at 4 °C. Seedlings were grown in a growth chamber under controlled conditions.

Generation of transgenic lines

The sequences for LSM4^R and LSM4^RxK were synthesized and cloned into the pDONR/Zeo (Invitrogen) via the Gateway method. The LSM4^RxK sequence was generated by changing the R codons to K codons in the nine methylable RGG domains. The LSM4 overexpression vectors (under CaMV 35S) were generated by introducing the AtLSM4^R or AtLSM4^RxK sequence into the binary vector pEarleyGate 100 by LR recombination. For overexpressor lines with YFP N-terminal tag, the pEntry vectors were recombined to pEearleyGate104. We identified the 885-bp sequence upstream of the LSM4 start codon as its promoter. Overlapping PCRs were done to generate the cassette containing LSM4 promoter driving the expression of FLAG-cDNALSM4 fusion protein. LSM4 cDNAs version were amplified from the above-mentioned vectors. This sequence was cloned in the pGWB516 vector by restriction with XbaI and SacI sites, to generate LSM4:FLAG-LSM4^R and LSM4:FLAG-LSM4^RxK. All plasmids were then introduced into Agrobacterium tumefaciens GV3101 strain and transformed into lsm4-1 heterozygous plants using the floral dip method (Clough and Bent 1998). Transgenic lines were selected on MS media supplemented with 15 mg/L phosphinotricin (Duchefa) or hygromicyn and genotyped for the lsm4-1 mutation using primers described in Supplementary Data Set 1. After genotyping, homozygous lsm4-1 plants bearing the different transgenes (T1 plants) were selfed and progeny selected that were homozygotic for the transgene. T3 lines were used for each study.

Circadian leaf movement analysis

For leaf movement analysis, seeds were sown into small circular pots (filled with growth media) and entrained in LD (16 h light/8 h dark) conditions at 22 °C for 7 d until the first pair of leaves were fully expanded. Then, plants were transferred to continuous light (LL) and constant temperature (22 °C) conditions. Photos were taken every hour for 6 d and analyzed by recording the plants' vertical leaf motion (relative vertical motion: RLM) with a program developed in java (Iserte and Yanovsky 2017) based on the Matlab code from the software TRiP: Tracking Rhythms in Plants (https://github.com/KTgreenham/TRiP) (Greenham et al. 2015). The circadian period was obtained via fast Fourier transform nonlinear least-squares (FFT-NLLS) analysis using the online program BioDare2 (www.biodare2.ed.ac.uk) (Zielinski et al. 2014). The first 24 h were excluded from the analysis to remove potential noise caused by the transfer from entrainment conditions to LL conditions.

RNA extraction and reverse transcription-quantitative PCR

Total RNA was isolated from seedlings using TRIZOL (Ambion) and treated with RNAse-free DNase I (Promega) to remove residual genomic DNA. One microgram of total RNA was used for reverse transcription (RT; Superscript II, Invitrogen). Transcript levels were quantified by RT-qPCR in a Stratagene MX3005P instrument (Agilent Technologies) using PP2A (AT1G13320) or IPP2 (AT3G02780) as the housekeeping gene. The sequences of the primers used to quantify the expression are listed in Supplementary Data Set 11.

Validation of splicing events

cDNAs were synthesized as above but with SuperScript (SS) III (Thermofischer). PCR amplification was performed using SuperFidelity Taq Enzyme (Invitrogen). Primers used for amplification are detailed in Supplementary Data Set 11. RT-PCR products were incubated with SYBR Green before electrophoresis on 2.5% (w/v) agarose gels. We selected the events to be validated according to the following parameters: the gene must have at least 50 reads to ensure good expression, the difference between the isoforms must not be larger than 10% in molecular weight, and when possible, the neighboring intron, with no changes, must be included in the PCR product to show that the measured event is specific. Unspliced ratio (Unspliced isoforms/Total isoforms) was determined by quantifying the bands of the unspliced and spliced isoforms of the gel with the Gel Analysis Tool of Image J.

RNA-seq library construction and sequencing

For RNA-seq experiments, WT, lsm4-1, LSM4^R, and LSM4^RxK in lsm4-1 background seedlings were grown on MS medium containing 0.8% (w/v) agar for 12 d under continuous light (LL) at 22 °C. Three biological replicates were collected, whole seedlings were harvested and total RNA was extracted with RNeasy Plant Mini Kit (QIAGEN) following the manufacturer's protocols. To estimate RNA concentration NanoDrop 2000c (Thermo Scientific) was used. Libraries were prepared and sequenced at the Max Planck-Genome-Centre Cologne (MP-GC).

RIP-seq analysis

RIP was done as previously described except for the bioinformatic analysis (Meyer et al. 2017; Köster and Staiger 2021). We calculated the enrichment of the signal from IP using 35S:YFP-LSM4 plants over an IP from the negative control plants 35S-YFP. Briefly, plants grown in MS plates in LL for 12 d were vacuum-infiltrated with 1% (v/v) formaldehyde for 15 min followed by quenching with 125 mm glycine. A whole-cell extract was prepared in RIP-lysis buffer. The extract was precleared with Sepharose beads and subjected to immunoprecipitation with GFP-Trap beads (Chromotek). After extensive washing with RIP washing buffer, co-precipitated RNAs were extracted with Trizol (Invitrogen) and treated with DNase (Promega). Libraries were prepared and sequenced at the MP-GC.

AS and differential expression analysis

AS analysis was performed as previously described in Mateos et al. (2023) using R with the ASpli package (Mancini et al. 2021). For expression analysis, genes with a FDR < 0.05 and log₂FC > |1.0| were considered as differentially expressed (DEGs) between genotypes. For splicing analysis, PSI and PIR were calculated and differential splicing was considered for bins with FDR < 0.15 and ΔPSI/PIR > 0.05. For the analysis of the effect of methylation we used a ΔPSI/PIR > 0.03. For the analyses of RNA-seq data from lsm8, bam files from our previous work (Carrasco-López et al. 2017) were treated as described above to compare splicing events.

Physiological response to ABA and salinity

Seeds were sown on MS medium supplemented with 0.5 or 1 µM ABA. The proportion of germinated seeds was scored after 48 h while greening after 7 d of treatment. Both parameters correspond to radicle emergence and fully opened cotyledons, 0.5 and 1 stages, respectively, according to Boyes et al. (2001). For the analysis of PR elongation, LRs formation, and FW, 4-d-old seedlings were grown vertically on MS agar plates and then transferred to MS supplemented with indicated concentrations of ABA, NaCl or sorbitol. Quantifications were assayed after 7 d. Chlorophyll content was measured spectrophotometrically at 645 and 663 nm according to Arnon (1949).

Infection assay

Five-week-old plants were inoculated into the abaxial side of eighth to tenth leaves with a bacterial suspension of P. syringae pv. tomato DC3000 (OD600 = 0.0002) or mock solution (10 mm MgSO₄). Three disks were punched from each leaf 48 hpi and bacterial growth was determined by counting bacterial colonies in plate assays according to de Leone et al. (2020).

Methylation status and PR level

Seven-days-old seedlings were transferred to liquid MS medium supplemented with 10 µm ABA for indicated times and assayed for the LSM4 methylation status by anti-SYM10 immunoblot. Five-week-old plants were inoculated with a bacterial suspension of P. syringae pv. tomato DC3000 (OD600 = 0.002) or mock solution (10 mm MgSO₄), harvested 4, 24, and 48 hpi and assayed for LSM4 methylation status and PR induction by immunoblots. Total proteins were extracted in extraction buffer (50 mm Tris–HCl pH 7.2, 100 mm NaCl, 10% (v/v) glycerol, 0.1% (v/v) Tween-20, 1 mm phenylmethylsulfonyl fluoride (PMSF), and plant protease inhibitor (Sigma)) and then centrifuged at 12,000 × g and 4 °C for 15 min. Equal amounts of protein (36 µg) were loaded onto SDS–PAGE, quantified by Bradford Protein assay, electro-transferred onto nitrocellulose membranes and probed with anti-SYM10 (Dilution 1/1,000) (Millipore, 07-412), anti-PR1 (Dilution 1/1,000) (Agrisera, AS10 687), or anti-PR2 (Dilution 1/1,000) (Agrisera, AS12 2366) antibodies overnight followed by secondary antirabbit antibody coupled to peroxidase (Dilution 1/5,000) (Invitrogen 31460). Proteins were visualized using the ECL kit (Pierce, USA) in an ImageQuant LAS 4000, GE Healthcare. Ponceau staining is shown as loading control.

IP and methylation status of LSM4-GFP and LSM8-GFP fusion proteins

Protein was isolated from 2-wk-old 35S:LSM4-GFP or 35S:LSM8-GFP transgenic lines in IP buffer containing 50 mm Tris–HCl, pH 7.5, 150 mm NaCl, 10% (v/v) glycerol, 0.1% (v/v) Nonidet P40; 1 mm PMSF, and protease inhibitor cocktail (Sigma–Aldrich, USA). After centrifugation at 12,000 × g and 4 °C for 15 min, protein extracts were incubated with GFP-Trap Agarose (Chromotek, Germany) at 4 °C for 1 h with gentle rotation. After centrifugation at 2,500 × g for 4 min, an aliquot of the supernatant was saved for flow-through control. Beads were washed four times in IP buffer and bound proteins were extracted in 2.5X Laemmi SDS-sample buffer. R-methylated LSM proteins were identified by immunoblot with anti-SYM10 as previously described above.

Statistical analysis

Statistical analysis was conducted as described in the text and figure legends. Statistical data are provided in Supplementary Data Set 12.

Accession numbers

Gene sequence data in this article can be found in GenBank databases using accession numbers AT5G27720 (LSM4), AT4G31120 (PRMT5), AT1G56520, AT1G72900 (TIR-NBS7),AT1G13320 (PP2A), AT3G02780 (IPP2), AT1G45249 (ABF2) AT1G72770 (HAB1), AT2G01210 (ZAR1).

The RIP-seq and RNA-seq raw data have been deposited in ArrayExpress (Kolesnikov et al. 2015) at EMBL-EBI (www.ebi.ac.uk/arrayexpress), under accession numbers E-MTAB-12369 (RIP-seq dataset) and E-MTAB-12370 (RNA-seq dataset).

Acknowledgments

We thank Dr. Marlene Reichel for critical reading of the manuscript. We thank Kristina Neudorf, Elisabeth Klemme, and Melania Manzano for technical assistance and Martin Lewinski for bioinformatic assistance.

Author contributions

J.L.M., M.J.I., D.S., and M.J.Y. designed the research; Y.C.A., M.J.I., S.P.S., M.de.L., R.C., T.K., and J.L.M. performed the experiments, J.L.M. performed the bioinformatic analysis, J.L.M., M.J.I., and S.P.S. wrote the paper.

Supplementary data

The following materials are available in the online version of this article.

Supplementary Figure S1. Bioinformatic analysis of 3′ ss and 5′ ss sequences.

Supplementary Figure S2. Direct binding of LSM4 to transcripts alters both AS and mRNA levels.

Supplementary Figure S3. Molecular complementation of LSM4^R and LSM4^RxK on splicing defects caused by LSM4 mutation.

Supplementary Figure S4. LSM4 methylation under abiotic stress.

Supplementary Figure S5. LSM4 affects splicing of abiotic stress related genes.

Supplementary Figure S6. Effect of LSM4, PRMT5, and LSM4 methylation on splicing of abiotic stress related genes.

Supplementary Figure S7. Representative models and retained intron splice forms of the five TIR-NBS-LRR genes.

Supplementary Figure S8.LSM4 transcript levels in transgenic lines.

Supplementary Figure S9.PR1 transcript levels in transgenic lines.

Supplementary Figure S10. Expression levels of LSM4 direct targets.

Supplementary Figure S11. LSM8-GFP is not methylated at its arginines.

Supplementary Figure S12. PRMT5 effect on LSM4 regulated genes.

Supplementary Figure S13. LSM4^R and LSM4^RxK expressed under LSM4 promoter complement physiological and molecular lsm4-1 phenotype under normal growth conditions.

Supplementary Data Set 1. List of enriched genes of RIP-seq of 35S:YFP-LSM4 in lsm4-1 compared to 35S-YFP plants.

Supplementary Data Set 2. List of AS events altered in lsm4-1 compared to WT plants.

Supplementary Data Set 3. List of DEGs in lsm4-1 compared to WT plants.

Supplementary Data Set 4. List of AS events altered in prmt5-5 compared to WT plants.

Supplementary Data Set 5. List of DEGs in LSM4^R in lsm4-1 vs lsm4-1 plants.

Supplementary Data Set 6. List of DEGs in LSM4^RxK in lsm4-1 vs lsm4-1 plants.

Supplementary Data Set 7. List of AS events altered in LSM4^Rin lsm4-1 and lsm4-1 compared to WT plants.

Supplementary Data Set 8. List of AS events altered in LSM4^RxK in lsm4-1 and lsm4-1 compared to WT plants.

Supplementary Data Set 9. List of genes differentially splicing among LSM4^R and LSM4^RxK plants.

Supplementary Data Set 10. Genes affected by LSM4 methylation status that are mis-spliced upon bacterial infection by Golisz et al. (2021).

Supplementary Data Set 11. List of primers used in this study.

Supplementary Data Set 12. Results from statistical test used in this study.

Funding

J.L.M. was supported by the Alexander von Humboldt-Stiftung (Alumni Program), Prestamo BID PICT2010-0641 from ANPCyT and the Max-Planck Society (Partner Group Program). M.J.I., M.J.Y., S.P.S., M.de.L., and J.L.M. were supported by the Consejo Nacional de Investigaciones Cienttíficas y Técnicas, Argentina (CONICET). Y.A. was supported by the ANPCyT. M.J.I. was supported by the Prestamo BID PICT-2020-4354 from ANPCyT. D.S. was supported by a Bilateral Grant CONICET and DFG to J.L.M. and DS STA653/9-1.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• MYB59 Gramene: AT5G59780

• MYB59 Araport: AT5G59780

• SKIP2 Gramene: AT5G67250

• SKIP2 Araport: AT5G67250

• SMD3 Gramene: AT1G76300

• SMD3 Araport: AT1G76300

• ABF2 Gramene: At1g45249

• ABF2 Araport: At1g45249

• AHG2 Gramene: AT1G55870

• AHG2 Araport: AT1G55870

• GORK Gramene: AT5G37500

• GORK Araport: AT5G37500

• AT1G56520 Gramene: AT1G56520

• AT1G56520 Araport: AT1G56520

• ILL3 Gramene: AT5G54140

• ILL3 Araport: AT5G54140

• PRMT7 Gramene: AT4G16570

• PRMT7 Araport: AT4G16570

• HAB1 Gramene: AT1G72770

• HAB1 Araport: AT1G72770

• PP2A Gramene: AT1G13320

• PP2A Araport: AT1G13320

• PR1 Gramene: AT2G14610

• PR1 Araport: AT2G14610

• RD22 Gramene: AT5G25610

• RD22 Araport: AT5G25610

• LSM4 Gramene: AT5G27720

• LSM4 Araport: AT5G27720

• ETI Gramene: Effector-triggered immunity

• ETI Araport: Effector-triggered immunity

• PTI Gramene: pattern-triggered immunity

• PTI Araport: pattern-triggered immunity

• TIR-NBS7 Gramene: AT1G72900

• TIR-NBS7 Araport: AT1G72900

Author notes

Yamila Carla Agrofoglio and María José Iglesias contributed equally to this work.

Present address: School of Biological Sciences, University of Auckland, 3A Symonds St, Auckland, New Zealand.

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 (https://academic.oup.com/plcell/pages/General-Instructions) is Julieta L. Mateos ([email protected]).

Conflict of interest statement. None declared.