MAC5A dual-functionally regulates cold responses at both transcriptional and post-transcriptional levels

Graphical abstract

Highlights

• •MAC5A is a conserved cold-responsive RBP across species.
• •MAC5A positively regulates plant freezing tolerance.
• •Loss-of-function of MAC5A altered the cold-responsive transcriptomic landscape.
• •MAC5A orchestrates plant cold responses at both transcriptional and post-transcriptional levels.

Abstract

Introduction

Plant cold responses have been extensively studied at the transcriptional level, particularly focusing on conserved transcription factors (TFs) such as C-repeat binding factors (CBFs). However, the post-transcriptional roles of RNA-binding proteins (RBPs) remain largely unexplored, with evolutionarily conserved cold-responsive RBPs yet to be investigated across species.

Objectives

This study aimed to identify conserved cold–responsive RBPs across plant species and investigate how a candidate RBP, MAC5A, regulates cold tolerance.

Methods

Integrative analyses of multi-species comparative genomes and low-temperature transcriptomes identified conserved cold-responsive RBPs. MAC5A loss-of-function and overexpression lines were examined to validate its genetic role in cold tolerance. Comparative transcriptomics and CUT&Tag-seq were applied to determine MAC5A-modulated alternative splicing (AS) and cold-regulated (COR) gene expression, which were further validated through minigene and dual-luciferase assays.

Results

We identified multiple conserved cold-responsive RBPs and focused on Arabidopsis MAC5A, a spliceosome component. Cold-induced MAC5A expression was independent of both CBF regulation and ABA signaling. Loss-of-function of MAC5A decreased freezing tolerance, while its overexpression enhanced it. The mac5a mutant exhibited extensive cold-responsive transcriptome alterations, with thousands of genes showing differential AS. Minigene assays confirmed MAC5A modulates splicing of COR genes, including CCA1, PRR9, ABF3 and bHLH112. Additionally, MAC5A regulated expression of other COR genes such as ABI1, COR28, GolS2, RD20, GRP4, DIN10 and NIA1 at the transcriptional level, likely through protein complexes. Notably, limited overlap existed between MAC5A-modulated AS genes and transcriptionally regulated COR genes, suggesting largely independent regulatory pathways.

Conclusions

MAC5A functions as a dual regulator to orchestrate cold response through both transcriptional and post-transcriptional control of COR genes. The limited overlap between genes regulated by these two layers indicates potential uncoupled regulatory manners, warranting further investigation into their coordination during cold stress.

Introduction

Low temperature impacts plant growth and development, restricting species' natural geographical distribution. Despite their inherent immobility, plants are not as completely passive as they seem. Over the long course of evolution, many plants have evolved sophisticated abilities to withstand or adapt to cold stress [1]. One such mechanism is cold acclimation, which allows temperate plants to acquire freezing tolerance and survive winter [2]. Cold acclimation is a dynamic and multifaceted process involving coordinated regulatory networks. Cold stress triggers transcriptional reprogramming and causes a cascade of metabolic and cellular modifications, leading to organism-wide structural adaptations and phenotypic plasticity [3]. These adaptive changes are largely mediated by cold-regulated (COR) genes, whose protein products play pivotal roles in conferring cold tolerance [4].

The regulators of COR genes in plant cold-responsive signaling pathways have been widely identified [5], among which C-repeat binding factors (CBFs) are the best-characterized [6], [7]. CBFs specifically target the CRT/DRE elements within the promoters of COR genes [8], [9], [10]. Since their discovery, CBF orthologs have been identified in numerous plant species, demonstrating the conservation of the CBF-dependent cold regulatory pathway across many plant species [11], [12], [13].

Despite the key roles of CBFs, increasing evidence suggests that CBFs directly regulated only ∼10% of COR genes [7], [8], [9], [14], suggesting the involvement of additional regulators in plant cold responses. Park et al. identified a set of transcription factors (TFs) such as HSFC1, ZAT12 and CZF1 that cooperate with CBFs to regulate COR genes [14]. Beyond the CBF pathway, regulators from various signaling pathways have been identified to modulate cold tolerance. For example, light signaling components, such as PIFs, participate in cold tolerance regulation [15]. Photomorphogenesis regulators like BBX7, BBX8 and BBX29 are also crucial for cold response [16], [17]. In addition, circadian clock components such as CCA1, LHY, PRR7, PRR9, RVE4 and RVE8, act as critical regulators of cold tolerance [18], [19], [20]. Moreover, brassinosteroid signaling TFs such as BZR1, BES1 and CESTA enhance cold tolerance [21], [22], while ethylene signaling TFs like EIN3/EIL1 negatively regulate it [23]. These findings highlight the complexity of TF regulation in cold responses and tolerance.

Beyond transcriptional regulation, post-transcriptional modulation plays a critical role in plant cold responses [24], [25]. RNA binding proteins (RBPs) are pivotal post-transcriptional regulators that modulate alternative splicing (AS) and other RNA processes [26], [27]. Accumulating evidence demonstrates that RBP-modulated AS is important for plant cold responses [28], [29], [30]. For example, the spliceosome component SmEb regulates AS of COP1 and BES1, which are critical for cold tolerance [28]. The U1 snRNP component RBP45d directly regulates AS during temperature-mediated flowering [31]. In addition, splicing factors such as STABILIZED1, the spliceosome component LSM2-8 complex, and the chloroplast RBP CP29A, regulate cold responses through COR gene splicing [32], [33], [34]. The cold-inducible RNA helicase RCF1 is essential for pre-mRNA splicing and COR gene regulation under cold stress [35]. Recently, CBFs were shown to be involved in modulating AS by directly interacting with the spliceosome complex subunit SKIP [36].

Although numerous RBPs have been identified in plant cold response, most studies focus on single species with limited cross–species comparison, leaving the functional conservation of RBP homologs poorly understood. Recently, we identified a group of cold-responsive RBPs that were preferentially retained following whole-genome duplications coinciding with global cooling [24]. Among these, MAC5A, a multidomain RBP involved in various regulatory pathways [37], [38], [39], [40], emerged as a conserved cold-responsive RBP across species. However, the molecular mechanism by which MAC5A modulates cold response remains unknown.

In this study, we first examined whether cold induction of Arabidopsis MAC5A depends on CBF or abscisic acid (ABA) signaling pathways. We then compared freezing tolerance of the mac5a mutant and MAC5A overexpression plants with wild-type (Col-0), and analyzed transcriptomic changes between mac5a and Col-0 under cold stress (0, 2, and 24 h). We detected altered splicing of COR genes in mac5a and verified MAC5A's modulation of their alternative splicing. Besides its RNA-recognition motif (RRM), MAC5A contains a CCCH-type zinc-finger domain [37], [38], which can mediate protein binding to RNA, DNA, and proteins [41], [42]. We next investigated whether MAC5A regulates COR gene transcription and tested its DNA-binding capacity. Our findings reveal that MAC5A orchestrates plant cold response by modulating COR genes at both post-transcriptional and transcriptional levels.

Materials and methods

Identification of conserved cold-responsive RBP orthogroups across species

To identify conserved cold-responsive RBPs across species, we leveraged our previous phylotranscriptomic dataset in which representative angiosperm species were subjected to cold stress at four time points [43]. From this dataset, we focused on five representative eudicot species (Arabidopsis thaliana, Carya illinoinensis, Glycine max, Betula pendula and Populus trichocarpa) and retrieved their homologous RBPs using known Arabidopsis RBPs as queries [24]. Coding sequences were classified into orthogroups using OrthoFinder 2.3.8 with default parameters [44]. Cold-responsive orthogroups were identified by integrating RNA-seq data from these five species under cold stress, following the method described previously [17]. Briefly, COR genes were identified by two criteria: 1) gene expression value with a trimmed mean of M-value (TMM) > 20 across three independent replicates, and 2) > 2–fold expression change at any cold–treatment time point relative to control. Orthogroups containing at least one COR gene were considered cold-responsive, and those showing consistent cold induction across all five species were defined as conserved cold-responsive RBP orthogroups (Table S1).

Plant materials and growth condition

The mac5a, cbf123, abi1 and aba3 mutants are in the Arabidopsis ecotype Columbia (Col-0) background, and Col-0 was utilized as the wild-type control. Seeds were sown on 1/2 Murashige & Skoog (MS) medium and treated at 4℃ in the dark for 3 days (d). Seedlings were then grown at 22°C in an artificial chamber under a 12/12 h light/dark photoperiod for 10 d. Then, the seedlings were transplanted into polythene pots with soil for further cultivation.

Freezing tolerance assay

Full length of MAC5A coding sequence was cloned into the pCAMBIA1300-Myc vector, and MAC5A overexpression plants (35S::MAC5A) were generated using the Agrobacterium-mediated genetic transformation. Three-week-old mac5a mutant, 35S::MAC5A, and wild-type Col-0 seedlings in similar growth status were selected for assays. Seedlings of each line were divided into four groups: two groups were cold-acclimated (4℃ for 3 d) before freezing treatment, while the other two groups were directly subjected to freezing. The freezing procedure was conducted as previously described [17], [45]. Three independent freezing treatments were conducted, and at least six technical replicates were performed for each experiment.

RNA-seq experiments and analysis

Three-week-old mac5a mutant and wild-type Col-0 seedlings cultured at 22°C under a 12/12 h light/dark photoperiod were treated with cold stress (4°C) for 0 h, 2 h, and 24 h, respectively. Total RNA from each sample was extracted for generating cDNA library and high-throughput sequencing on Illumina NovaSeq 6000 (Annoroad, China). Three biological replicates were performed for each treatment. Transcriptomic analysis was conducted as previously described [46]. CASH was utilized to detect differential alternative splicing (DAS) events (adjusted P value ≤ 0.01) between the compared samples [47]. Differentially expressed genes (DEGs) were obtained by false discovery rate (FDR) ≤ 0.05 and absolute fold change ≥ 2.

Minigene assay

DNA fragments with 'exon–intron-exon' structure from MAC5A putative regulated genes were cloned into pCAMBIA1300 to construct minigenes (Gene-mini). Meanwhile, the MAC5A coding sequence was cloned into pK2GW7-eYGFP to generate MAC5A-eYGFP. Agrobacterium tumefaciens strains harboring minigene and MAC5A-eYGFP constructs were co-transfected into N. benthamiana leaves. After 72 h of cultivation, the injected region of N. benthamiana leaves were collected and total RNA was extracted from each sample. Expression of AS fragment that exhibited intron retention was detected by qRT-PCR, and the relative expression of each AS fragment was normalized to eYGFP. Three independent experiments were conducted, and primers utilized for minigene assay were provided in Table S6.

CUT&Tag sequencing and data analysis

Two-week-old 35S::MAC5A overexpression and wild-type (Col-0) seedlings grown on 1/2 MS medium were utilized for protoplasts isolation. CUT&Tag assays of 35S::MAC5A and Col-0 were performed as previously described [17]. Three independent CUT&Tag replicates were performed on 35S::MAC5A plants using an anti-Myc antibody (35S::MAC5A_anti-Myc). Two independent controls [Col-0_anti-Myc (Control-1) and 35S::MAC5A_anti-IgG (Control-2)] were also performed. Extracted DNA fragments were subjected to high-throughput sequencing on Illumina NovaSeq 6000 (Annoroad, China). We performed peak calling between 35S::MAC5A_anti-Myc samples and controls using MACS v2.7.1 [48] and annotated peaks to their nearest genes by ChIPseeker v1.20 [49]. Replicate concordance was evaluated by quantifying the overlap percentage of peaks between different biological replicates of 35S::MAC5A. The shared peaks among replicates were utilized for further analysis.

Dual-luciferase assay

To generate effector (35S::MAC5A) and reporter (Promoter-LUC) constructs for dual-luciferase assays, MAC5A coding sequence and promoters of its modulated genes were cloned and inserted into pGreenⅡ 62-SK and pGreenⅡ 0800-Luc, respectively. Recombinant vectors and negative control vectors [pGreenⅡ 62-SK (35S::Empty) and pGreenⅡ 0800-Luc (35S::LUC)] were transformed into Agrobacterium tumefaciens strains GV3101 (pSoup-p19). A. tumefaciens strains harboring effector and reporter constructs were co-transformed into N. benthamiana leaves. After 72 h of cultivation, the injected leaves were collected and incubated with 150 μg/mL D-Luciferin potassium salt solution (Beyotime, China). Luciferase activity was detected on a Tannon 5200 Chemiluminescence imaging system (Tanon, China). At least three biological replicates were performed. Primers used for generating reconstructed vectors were provided in Table S12.

Yeast one-hybrid assay

DNA sequences of MAC5A-associated binding peaks were cloned into pAbAi vector to construct pBait-AbAi plasmids. Meanwhile, MAC5A coding sequence was cloned into pGADT7 AD vector to construct prey plasmid. Each pBait-AbAi plasmid was transformed into Y1Hgold yeast strains, and minimal Aureobasidin A (AbA) concentration that inhibit pBait-AbAi autoactivation was determined on SD/-Ura/AbA medium. Subsequently, prey plasmid was separately transformed into the pBait-AbAi strains for yeast one-hybrid (Y1H) assay, with colonies selected on SD/-Leu/AbA medium.

Electrophoretic mobility shift assay

MAC5A coding sequence was cloned into pGEX-4T-1 vector and expressed in Escherichia coli BL21. MAC5A-GST fusion protein was purified by GST-tag Protein Purification Kit (Beyotime, China). Meanwhile, two enriched motifs around MAC5A associated binding regions and a corresponding mutant motif were labeled with biotin using EMSA Probe Biotin Labeling Kit (Beyotime, China). Finally, interaction between MAC5A-GST protein and the labeled/unlabeled motifs was evaluated by electrophoretic mobility shift assay (EMSA) with Chemiluminescent EMSA Kit (Beyotime, China).

Results

20 conserved cold-responsive RBP orthogroups are identified across five species

Based on integrative analysis of comparative genomics and low-temperature transcriptomics (see Methods), we identified 20 conserved cold-responsive RBP orthogroups across five eudicot species (Fig. 1A, Table S1). Among these orthogroups, some have previously been shown to play roles in temperature responses. For example, GRP2 and GR-RBP4 within the RBG-1 orthogroup are cold–inducible, and GRP2 is required for enhanced cold and freezing tolerance in Arabidopsis [50], [51]. HSP70-3, a member of the HSP70 orthogroup, interacts with the RNA binding protein SGS3 to co-regulate transgenerational thermomemory [52]. Additional RBPs in the HSP90, NUCL, PTB, ECT, and RH–1/2/3 orthogroups also participate in diverse RNA-binding processes [53], [54], [55], [56], [57]. Notably, the MAC5 orthogroup includes MAC5A, a multidomain spliceosome component involved in various pathways [37], [38], [39], [40]. Here, we selected MAC5A as a representative of the conserved cold-responsive RBPs to investigate its role in plant cold response (Fig. S1).

Fig. 1.

Cold induction of MAC5A is CBF- and ABA-independent. (A) Conserved cold-responsive RBP genes across five eudicot species: Arabidopsis thaliana (Ath), Betula pendula (Bpe), Carya illinoinensis (Cil), Glycine max (Gma), and Populus trichocarpa (Ptr). RBP orthogroup genes and their expression under normal (0 h) and cold stress (2, 24, and 168 h) conditions were obtained by previous studies (see Methods) [17], [24]. Cold induction was defined as FDR < 0.01 and fold change > 2 at any timepoint (2, 24, or 168 h) of cold stress (4℃). The MAC5 orthogroup containing MAC5A is highlighted in red. (B-C) MAC5A expression is CBF-independent. No CBF1/2/3 ChIP-seq binding peaks were detected across the Arabidopsis MAC5A locus, and MAC5A RNA-seq expression levels remained unchanged in the Arabidopsis cbf123 mutant under cold stress (B). The ChIP-seq and RNA-seq data were re-analyzed from publicly available datasets [7], [9]. This analysis was further validated by our qRT-PCR experiment (C). (D) MAC5A expression is ABA-independent. The qRT-PCR result of MAC5A expression in the ABA-insensitive mutant abi1 and the ABA-deficient mutant aba3 of Arabidopsis. Each qRT-PCR experiment was performed with three independent biological replicates, and the relative expression level of MAC5A was normalized to ACTIN2 (AT3G18780) and ACTIN8 (AT1G49240). Primers used for qRT-PCR were listed in Table S3. Significant difference of gene expression between two treatments or samples were determined by a minimum two-fold change (> 2) and a statistical P value (P < 0.05) (Student's t-test).

Cold induction of MAC5A is independent of CBF and ABA signaling

CBFs are the best-known transcription factors in plant cold acclimation and are highly conserved in angiosperms [58]. To determine whether cold induction of MAC5A is directly regulated by CBF, we examined CBF1/2/3 binding profiles across the MAC5A locus using published ChIP-seq data [7]. No CBF1/2/3 binding peaks were detected (Fig. 1B). Additionally, RNA-seq analysis showed that loss-of-function of CBFs (cbf123 mutant) did not alter the cold-induction of MAC5A (Fig. 1B). This result was further confirmed by qRT-PCR (Fig. 1C). Since many COR genes are induced in an ABA-dependent manner [59], [60], we next examined whether cold induction of MAC5A is linked to ABA signaling. qRT-PCR analysis showed similar MAC5A expression patterns in Col-0 and both the aba3 and abi1 mutants (Fig. 1D), indicating that cold-indued regulation of MAC5A occurs via a CBF-independent and ABA-independent pathway.

Effects of MAC5A loss-of-function and overexpression on freezing tolerance

To explore the role of MAC5A under cold stress, we generated 35S::MAC5A overexpression lines in Arabidopsis (Fig. S2) and assessed the effects of MAC5A loss-of-function and overexpression on freezing tolerance. Under normal conditions, wild-type Col-0, mac5a mutant, and 35S::MAC5A plants displayed healthy and comparable growth during the seedling stage. However, following freezing treatment, mac5a showed hypersensitivity compared to Col-0, while 35S::MAC5A exhibited enhanced freezing tolerance (Fig. 2A). Consistent with these observations, survival rates were significantly higher in 35S::MAC5A and lower in mac5a compared to Col-0 (Fig. 2B). These results suggest that MAC5A acts as a positive regulator of freezing tolerance in Arabidopsis.

Fig. 2.

Effect of loss-of-function and overexpression of MAC5A on freezing tolerance. (A) Phenotypes of the Arabidopsis Col-0, mac5a mutant and 35S::MAC5A overexpression plants under freezing treatments with or without cold acclimation. Plants treated with cold acclimation were prior exposure to non-lethal low temperature (4℃) for 3 days before freezing treatment. Photographs were taken after 3 d of recovery at 22℃ under a 12/12 h light/dark photoperiod. (B) Survival rate of Col-0, mac5a and 35S::MAC5A after freezing tolerance assays. Survival rate of each line was calculated after 5 d of recovery at 22℃. At least three independent experiments were performed, and at least six biological replicates were utilized for each experiment. Asterisks represent a significant difference between Col-0 and mac5a/35S::MAC5A (**P < 0.01, Student's t-test).

Since MAC5A was identified as a conserved cold-responsive RBP across eudicot species (Fig. 1A), we next examined whether MAC5A homologs in monocots, such as Phyllostachys edulis and Oryza sativa, were also cold responsive. Both PheMAC5A and OsMAC5A were cold-inducible, and their overexpression in Arabidopsis conferred enhanced freezing tolerance (Fig. S3). These findings suggest that the role of MAC5A in cold responses may be conserved between eudicots and monocots, supporting a potentially conserved function across angiosperms.

Loss-of-function of MAC5A alters alternative splicing of over 1,100 genes

To determine MAC5A's regulatory function in plant cold response, we compared transcriptome-wide differential alternative splicing events (DASEs) between mac5a mutant and Col-0 under cold stress. MAC5A-affected cold-responsive DASEs were determined based on two criteria: (1) significantly differential AS in cold-treated Col-0 compared to non-treated controls, and (2) significantly differential AS between mac5a and Col-0 under cold stress (see Methods). We identified 1,371 MAC5A-affected cold-responsive DASEs across 1,120 differentially alternative spliced genes (DASGs) (Fig. 3A, Table S4). Analysis of AS types revealed that intron retention (IR) was most prevalent, followed by exon skipping (ES) (Fig. 3B).

Fig. 3.

Loss-of-function of MAC5A alters alternative splicing landscape under cold stress. (A) Comparative analysis of DASEs and DASGs in Col-0 and mac5a under cold stress. Numbers above the parentheses indicate DASEs, while those inside the parentheses represent DASGs. DASEs are defined with FDR < 0.01 at 2 or 24 h of cold stress (4℃) and DASGs are defined as genes containing at least one DASE (see Methods). (B) Overview of five AS types differentially regulated in mac5a under cold stress. (C) Gene ontology (GO) enrichment terms of the 1,120 DASGs affected by MAC5A. (D) Venn diagram of the overlap between MAC5A-affected DASGs and COR genes.

Gene ontology (GO) analysis showed that MAC5A-affected DASEs were enriched in RNA splicing, mRNA stability, cold acclimation, DNA repair, and reproductive development (Fig. 3C, Table S5). GO terms related to circadian rhythm, light stimulus, abiotic stress, and gene expression regulation were also significantly enriched. These findings indicate a multifaceted role of MAC5A in RNA splicing, plant development, and abiotic stress responses, including cold stress. To detect whether MAC5A-affected DASGs were also transcriptionally cold-responsive, we compared the 1,120 DASGs with COR genes identified in Col-0. The analysis revealed that 454 (∼40%) MAC5A-affected DASGs were cold-regulated at both the transcriptional and splicing levels (Fig. 3D).

MAC5A modulates alternative splicing of genes in cold response

To validate whether AS events were modulated by MAC5A, we selected four representative MAC5A-affected DASGs, such as CCA1, PRR9, ABF3, and bHLH112, and visualized their AS regions using RNA-seq data (Fig. 4A). qRT-PCR analyses confirmed significant accumulation of specific AS regions in the mac5a mutant (Fig. 4B), suggesting that loss of MAC5A function alters AS events in these genes under cold stress. Minigene assays further demonstrated that MAC5A inhibits the production of these AS fragments of CCA1, PRR9, ABF3 and bHLH112 (Fig. 4C-D), consistent with their accumulation in the mac5a mutant (Fig. 4A-B). Given that these genes have well-established functions in cold responses [61], [62], [63], MAC5A-mediated AS regulation likely contributes to the altered freezing tolerance observed in mac5a and 35S::MAC5A plants.

Fig. 4.

MAC5A modulates alternative splicing of COR genes in cold response. (A) Four exemplified genes with differential AS regions in mac5a. The AS regions of CCA1, PRR9, ABF3 and bHLH112 were visualized using RNA-seq data. The rectangular area indicates the differential AS region between mac5a and Col-0. (B) Validation of the AS changes for the exemplified genes in mac5a. The AS fragment corresponds to the differential AS region indicated by the rectangle in (A). The relative expression levels of AS fragments were normalized to ACTIN2 (AT3G18780) and ACTIN8 (AT1G49240). (C) Reconstructed vectors for the minigene assay. DNA fragments structured as 'exon–intron-exon' from the AS regions of CCA1, PRR9, ABF3 and bHLH112 in (A) were cloned and individually inserted into the pCAMBIA1300 vector to construct Gene-mini. Meanwhile, the full-length CDS of MAC5A was cloned and inserted into the pK2GW7-eYGFP vector to generate MAC5A-eYGFP. The pK2GW7-eYGFP vector, referred to eYGFP, served as a negative control for MAC5A-eYGFP. (D) Relative expression levels of the AS fragments in minigene assays. Agrobacterium strains containing the Gene-mini and either MAC5A-eYGFP or eYGFP vectors were co-transfected into N. benthamiana leaves. Relative expression level of the AS fragment generated by intron retention was detected by qRT-PCR after 72 h of co-transfection. The relative expression level of each AS fragment was normalized to eYGFP. At least three independent experiments were performed. Asterisks represent a significant difference between the two comparison groups (*P < 0.05, **P < 0.01, Student's t-test).

Loss-of-function of MAC5A impacts COR gene expression as well

In addition to RRM, MAC5A contains a CCCH-type zinc-finger domain, which has the potential to bind DNA and regulate expression of COR genes [41], [42]. This led us to investigate whether MAC5A also regulates COR gene transcription. We found that loss-of-function of MAC5A significantly altered the transcriptional profiles of COR genes (Fig. 5A). MAC5A-affected COR genes were determined based on two criteria: (1) significantly differentially expressed in Col-0 under cold stress compared to non-treated controls, and (2) significantly differentially expressed between mac5a and Col-0 under cold stress (see Methods). Using these criteria, we identified 575 MAC5A-affected COR genes (Fig. 5B; Table S7).

Fig. 5.

Loss-of-function of MAC5A impacts the transcription profiles of COR genes in cold response. (A) Transcription profiles of Col-0 and mac5a under normal and cold stress conditions. (B) Comparison of DEGs between Col-0 and mac5a under cold stress. (C) Significantly enriched GO terms of MAC5A-affected CORs. (D) Venn diagram of the overlap among MAC5A-affected CORs, CBFs-targeted CORs, and ABA-responsive CORs. CBFs-targeted CORs and ABA-responsive genes were obtained from the publicly available datasets [7], [64]. ABA-responsive CORs were identified by overlapping ABA-responsive genes and COR genes in Col-0. Genes in each of the groups are listed in Table S8. (E) Overlap between MAC5A-affected CORs and MAC5A-affected DASGs.

As expected, GO analysis revealed that these genes were significantly enriched in processes related to stress and hormone responses, plant growth and development, and signal transduction (Fig. 5C). Although cold induction of MAC5A is independent of CBF and ABA signaling pathways (Fig. 1), MAC5A may still co-regulate COR genes with these pathways. To test this, we compared MAC5A-affected COR genes with COR genes in the CBFs and ABA pathways [7], [64]. Venn diagram analysis showed a statistical overlap between MAC5A-affected CORs and ABA responsive CORs, but not with CBFs-targeted CORs (Fig. 5D). These findings suggest that MAC5A also functions as a transcription-associated regulator of COR genes, with MAC5A-affected COR genes being primarily CBF-independent but closely associated with the ABA signaling pathway. Notably, we observed minimal overlap between MAC5A-affected DASGs and MAC5A-affected CORs (Fig. 5E), indicating that MAC5A modulates plant cold response through largely independent regulation of alternative splicing and transcription levels of distinct gene sets.

MAC5A associates with chromatin to modulate COR gene expression

To examine whether MAC5A associates with chromatin and modulates COR genes on a genome-wide scale, we performed CUT&Tag-seq experiments using 35S::MAC5A overexpression plants. Compared to negative controls (anti-Myc for Col-0 and anti-IgG for 35S::MAC5A), thousands of chromatin regions co-precipitated with MAC5A under anti-Myc antibody incubation (Fig. 6A; Fig. S4; Table S9). Over 97% of the bound peaks were located in promoter regions (<3 kb upstream of TSS), with 78.5% within 1 kb (Fig. 6B). Motif analysis revealed significant enrichment of two motifs around MAC5A-associated peaks: a TC-rich motif and a GA-rich motif (Fig. 6C), which seem to represent complementary DNA strands.

Fig. 6.

Identification of MAC5A-associated binding peaks on a genome-wide scale. (A) Genome-wide overview of MAC5A-associated binding peaks. Control-1 represents the peaks identified in wild-type Col-0 using anti-Myc antibody, and Control-2 represents the peaks identified in 35S::MAC5A using anti-IgG antibody. (B) Distribution of MAC5A-associated binding peaks across different regions around genes. The pie indicates the proportion of peaks distributed in different annotated regions of genes. The UpSet plot shows intersections among MAC5A-associated binding peaks across genes. (C) Two enriched motifs around MAC5A-associated binding peaks. MEME-ChIP was used to discover the motifs [84].

To identify COR genes associated with MAC5A occupancy, we overlapped MAC5A-associated genes from CUT&Tag-seq with MAC5A-affected COR genes from RNA-seq. This comparison identified 175 COR genes that both associated with MAC5A and showed altered expression in the mac5a mutant (Fig. 7A; Table S10). GO analysis revealed that these genes were enriched in processes related to abiotic stress response, hormone response, and plant growth and development, including response to cold, response to abscisic acid, and regulation of seedling development (Fig. 7B). Among the 175 genes, several well-documented COR genes were identified, such as ABI1, COR28, GolS2, RD20, GRP4, DIN10 and NIA1 [65], [66], [67] (Fig. 7C; Table S10).

Fig. 7.

Identification of MAC5A-modulated COR genes. (A) Identification of MAC5A-modulated COR genes by overlapping MAC5A-associated genes and MAC5A-affected CORs through a Venn diagram. (B) GO enrichment terms of the 175 MAC5A-modulated COR genes. (C) A set of exemplified MAC5A-modulated COR genes. The MAC5A-associated binding peaks on these COR genes were visualized using CUT&Tag-seq data.

CUT&Tag-qPCR further confirmed the significant enrichment of MAC5A-associated binding peaks at these COR gene loci (Fig. 8A; Table S11), consistent with CUT&Tag-seq results (Fig. 7C). To examine the regulatory effects of MAC5A on these genes, we performed dual-luciferase assays and found that MAC5A negatively regulates expression of ABI1, COR28, GolS2, RD20 and GRP4, while positively regulating DIN10 and NIA1 (Fig. 8B-C). The regulatory effects of MAC5A on these genes align well with the transcriptional changes observed in the mac5a mutant (Fig. 8D; Table S7). Since ABI1 and COR28 have been reported to negatively regulate plant cold tolerance [65], [66], MAC5A-mediated repression of these genes may contribute to enhanced freezing tolerance in 35S::MAC5A plants compared to the mac5a mutant (Fig. 2). These results suggest that beyond its function in alternative splicing, MAC5A associates with chromatin at COR gene loci and regulates their expression at transcriptional level. Remarkably, Y1H and EMSA assays failed to detect direct DNA binding between MAC5A and its associated binding peaks (Fig. S5), indicating that MAC5A likely functions through protein co-regulatory complexes rather than direct DNA binding. Although our CUT&Tag-seq data defines the genome-wide occupancy of MAC5A, the epigenetic features of these loci (e.g., their overlap with active chromatin marks such as H3K4me3) and the recruitment partners of MAC5A remain to be determined.

Fig. 8.

Experimental validation of MAC5A's modulation on COR genes. (A) CUT&Tag-qPCR validation of the regulation of COR genes by MAC5A. The relative fold enrichment of each MAC5A-assocoated binding region was normalized to ACTIN2 (AT3G18780) and ACTIN8 (AT1G49240). At least three independent experiments were performed. Asterisks represent a significant difference between Col-0 and 35S::MAC5A (**P < 0.01, Student's t-test). (B) Schematic diagrams of effector and reporter constructs used for the dual-luciferase assay. The full-length CDS of MAC5A was inserted into the pGreenⅡ 62-SK vector to generate the effector vector (35S::MAC5A). The sequence of each MAC5A-assocaited binding region was inserted into pGreenⅡ 0800-Luc vector to generate the reporter vector (Promoter-LUC). The pGreenⅡ 62-SK (35S::Empty) and pGreenⅡ 0800-Luc (35S::LUC) vectors were separately utilized as negative controls of 35S::MAC5A and Promoter-LUC. (C) Detection of luciferase signals of each combination of effector and reporter in N. benthamiana. Detailed information of the four combinations was shown in the graphic resembling a leaf shape. At least three biological replicates were performed. (D) Comparison of the expression patterns of MAC5A-modulated COR genes under cold stress in mac5a mutant and the wild-type Col-0. Asterisks represent a significant difference between Col-0 and mac5a (fold change ≥ 2 and FDR ≤ 0.05). Gene expression value is represented by the average Trimmed Mean of M-value (TMM) of each gene from three independent replicates of our RNA-seq data (Table S7).

Discussion

Molecular mechanisms of cold response signaling have been widely characterized [5], yet most studies have focused on transcription factors. By comparing genomes and low-temperature transcriptomes across five eudicot species, we identified conserved cold-responsive RBP orthogroups (Fig. 1A, Table S1) and focused on the spliceosome component MAC5A [38]. MAC5A is a multidomain RBP involved in pri-miRNA processing, DNA damage response, and gibberellin-regulated stamen development [37], [38], [39], [40]. However, its role in cold stress remains uncharacterized. Here, we demonstrate that MAC5A is transcriptionally induced by cold and regulates cold responses through dual regulatory layers: post-transcriptionally by modulating alternative splicing and transcriptionally by associating with chromatin at COR gene loci. Our findings expand MAC5A's functional repertoire and reveal a dual-level regulatory mechanism at transcriptional and post-transcriptional regulation, highlighting multifaceted roles of RBPs in stress adaptation.

MAC5A appears to be a conserved regulator of cold tolerance

MAC5A is transcriptionally upregulated under cold stress via CBFs-independent and ABA-independent pathways (Fig. 1B-D), representing an alternative route in cold signaling networks. Freezing tolerance assays confirmed that MAC5A positively regulates cold tolerance (Fig. 2). Notably, MAC5A was identified as a conserved cold-responsive RBP across five eudicot species (Fig. 1A). Extending this observation, MAC5A homologs in monocots, including P. edulis and O. sativa, are also cold-induced and positively regulate cold tolerance in Arabidopsis (Fig. S3), suggesting its potential functional conservation across diverse plant lineages.

MAC5A regulates cold responses through AS of circadian genes

As a spliceosome component [37], [38], MAC5A regulates AS of COR genes under cold stress (Fig. 3A). GO enrichment demonstrated that these genes were associated with RNA splicing, cold acclimation, and circadian rhythm regulation (Fig. 3C). Loss-of-function of MAC5A caused AS changes in core circadian clock genes including CCA1, LHY, PRR9, RVE4, RVE7, RVE8, SICKLE (SIC), and EARLY-RESPONSIVE TO DEHYDRATION 7 (ERD7) [68], [69], [70], [71] (Table S4). Minigene assays confirmed MAC5A's regulatory effect on CCA1 and PRR9 splicing (Fig. 4D). This finding aligns with accumulating evidence that cold-induced AS of circadian genes plays crucial roles in temperature adaptation [61], [62], [68], [69], [70], [71], positioning MAC5A as a key regulator linking AS of circadian genes to cold tolerance. However, the minigene assay based on N. benthamiana may not fully recapitulate the physiological regulation of splicing under native conditions. Therefore, the detailed molecular mechanism by which MAC5A recognizes specific splice sites requires further investigation through RNA immunoprecipitation and high-resolution binding site mapping.

MAC5A regulates cold responses through chromatin-associated transcription

Beyond its post-transcriptional regulation, MAC5A also modulates COR gene expression at the transcriptional level (Fig. 5, Table S7). This dual functionality is consistent with the structure of MAC5A, which contains both RRM and CCCH-type zinc-finger domains [38]. RRM domains function in mRNA splicing [72], while CCCH-type zinc-finger proteins can regulate both RNA and DNA-level processes [41], [42], [73], [74].

CUT&Tag-seq revealed that MAC5A associates with promoter regions of thousands of genes, including 175 COR genes that showed altered expression in the mac5a mutant (Fig. 6, Fig. 7; Table S10). Over 97% of MAC5A-associated bound peaks localized within 3 kb upstream of TSS (Fig. 6B), with enrichment of GA-rich and TC-rich motifs (Fig. 6C). Moreover, CUT&Tag-qPCR and dual-luciferase assays confirmed that MAC5A modulates expression of well-documented COR genes, including ABI1, COR28, GolS2, RD20, GRP4, DIN10 and NIA1 (Fig. 8) [65], [66], [67]. Among the genes, ABI1 is a member of protein phosphatases 2C (PP2C) that negatively regulate ABA signaling. Antisense inhibition of PP2C could accelerate cold acclimation [65], implying a putative negative role of ABI1 in freezing tolerance. Additionally, the cor28 mutant exhibited enhanced freezing tolerance compared to the wild-type control, suggesting a negative role of COR28 in freezing tolerance [66]. MAC5A negatively regulates COR genes like ABI1 and COR28, explaining enhanced freezing tolerance in MAC5A overexpression plants compared to the mac5a mutant (Fig. 2).

The transcriptional changes in the mac5a mutant likely result from both indirect and direct effects. Indirectly, MAC5A may influence downstream transcription by modulating AS of transcription factors such as CCA1, PRR9, ABF3 and bHLH112 (Fig. 4). Directly, MAC5A may associate with chromatin and recruit transcription co-regulators to regulate COR gene expression.

MAC5A likely acts through protein complexes rather than direct DNA binding

Despite that CUT&Tag-seq and dual-luciferase assays suggested that MAC5A could modulate the transcription of COR genes, Y1H and EMSA assays did not detect direct DNA binding between MAC5A and its associated binding peaks (Fig. S5). This indicates that MAC5A functions through protein co-regulatory complexes rather than direct DNA binding. This mode of regulation is exemplified by SKIP, another spliceosome component that acts as both splicing factor and transcriptional regulator through distinct partners [75], [76], [77]. CDC5 functions in a similar manner, which controls both transcription and splicing of flowering genes [78]. MAC5A interacts with both CDC5 and SKIP [38], [75], as well as spliceosome components PRP8 and BRR2A [79], which have been proven to play roles in chromatin regulation [80], [81]. Motif analysis revealed enrichment of GA-rich elements similar to the binding motif of transcriptional regulator BPC1 [82], [83]. Therefore, the observed promoter activity changes of COR genes by MAC5A's modulation may depend on co-factor recruitment, chromatin remodeling, or both mechanisms. Whether MAC5A physically interacts with BPC1 or other GA-enriched motif binding factors remains an important question for future studies.

Limitations of this study

Our study reveals a paradigm in which MAC5A is involved in transcriptional and post-transcriptional regulation for cold tolerance. This dual-level regulatory mechanism provides new insights into how plants coordinate multiple layers of gene regulation under environmental stress. However, several questions remain to be addressed. First, the precise RNA-binding sites of MAC5A and the sequence determinants governing its substrate specificity require genome-wide mapping through techniques such as CLIP-seq or iCLIP. Second, identifying MAC5A's cold stress interactors (e.g., transcriptional regulators, chromatin modifiers) and determining how these complexes regulate chromatin states at COR genes will be essential for mechanistic clarity. Finally, understanding how these two regulatory modes are coordinated during cold stress represents an important direction for future investigation.

Conclusion

In summary, we identified 20 conserved cold-responsive RBP orthogroups and characterized MAC5A as a dual-level regulator of plant cold responses (Fig. 9). MAC5A is a CBF- and ABA-independent, cold-induced RBP that positively contributes to freezing tolerance through post-transcriptional and transcriptional layers. At the post-transcriptional level, MAC5A modulates AS of COR genes, particularly circadian clock components. At the transcriptional level, MAC5A associates with chromatin to regulate COR gene expression. Mechanistically, MAC5A does not directly bind DNA but likely functions through interactions with transcriptional co-regulators or chromatin-associated factors. While precise mapping of MAC5A RNA-binding sites and systematic analysis of its protein interactome remain to be determined, our findings establish that RBPs contribute to multilayered regulatory networks in plant stress responses, expanding our understanding of RNA-mediated cold-stress regulation in environmental adaptation.

Fig. 9.

Dual regulatory role of MAC5A in plant cold tolerance. A schematic model illustrating how MAC5A functions at both transcriptional and post-transcriptional levels under cold stress. MAC5A regulates alternative splicing of COR-related transcripts and modulates transcription of other COR genes through protein complexes, together enhancing plant cold tolerance.

Compliance with ethics requirements

This article does not contain any studies with human or animal subjects.

Data availability statement

RNA-seq data for mac5a mutant and Col-0, and CUT&Tag-seq data for 35S::MAC5A and Col-0, have been deposited at the National Genomics Data Center (PRJCA033595) and NCBI BioProject (PRJNA1198120).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: