Serrate-Associated Protein 1, a splicing-related protein, promotes miRNA biogenesis in Arabidopsis

Summary

• MicroRNAs (miRNAs) are essential regulators of gene expression in metazoans and plants. In plants, most miRNAs are generated from primary miRNA transcripts (pri-miRNAs), which are processed by the Dicer-like 1(DCL1) complex along with accessory proteins.

• Serrate-Associated Protein 1 (SEAP1), a conserved splicing-related protein, has been studied in human and yeast. However, the functions of SEAP1 in plants remain elusive.

• Lack of SEAP1 results in embryo lethality and knockdown of SEAP1 by an artificial miRNA (amiR^SEAP1) causes pleiotropic developmental defects and reduction in miRNA accumulation. SEAP1 associates with the DCL1 complex, and may promote the interaction of the DCL1 complexes with pri-miRNAs. SEAP1 also enhances pri-miRNA accumulation, but does not affect pri-miRNA transcription, suggesting it may indirectly or directly stabilize pri-miRNAs. In addition, SEAP1 affects the splicing of some pri-miRNAs and intron retention of messenger RNAs at global levels.

• Our findings uncover both conserved and novel functions of SEAP1 in plants. Besides the role as a splicing factor, SEPA1 may promote miRNA biogenesis by positively modulating pri-miRNA splicing, processing and/or stability.

Introduction

MicroRNAs (miRNAs) are 20–24 nucleotide (nt) regulatory RNAs that play essential roles in various biological processes such as development and stress responses in both plants and animals (Voinnet, 2009; Achkar et al., 2016; Cui et al., 2017; Song et al., 2019; Li & Yu, 2021). miRNAs mainly repress gene expression through directing messenger RNA (mRNA) degradation and/or translational repression (Baulcombe, 2004; Voinnet, 2009). Biogenesis of most miRNAs begins with the transcription of primary miRNA transcripts (pri-miRNAs) from miRNA encoding genes (MIRs) by the DNA-dependent RNA polymerase II (Pol II) (Xie et al., 2005; Wang, J et al., 2019b). In plants, pri-miRNAs are co-transcriptionally processed by the RNase III enzyme DICER-LIKE 1 (DCL1) to produce miRNAs in the nucleus. SERRATE (SE; a C2H2 zinc finger protein), HYPONASTIC LEAVES 1 (HYL1; a double-stranded protein) and TOUGH (an RNA-binding protein) form a complex with DCL1 to ensure efficient and accurate processing of pri-miRNAs (Lobbes et al., 2006; Yang et al., 2006; Dong et al., 2008; Ren et al., 2012; Zhang et al., 2015). The intrinsically disordered regions of SE are required for phase separation, which is crucial for the assembly of the DCL1 complex (Xie et al., 2021). After processing, miRNAs are 2’-O-methylated by HUA ENHANCER 1 (an RNA methyltransferase), incorporated into AGONAUTE 1 (AGO1) with the aid from HEAT SHOCK PROTEIN90 and CYCLOPHILIN40, to regulate gene expression (Baumberger & Baulcombe, 2005; Yu et al., 2005; Smith et al., 2009; Earley & Poethig, 2011; Bologna et al., 2018; Zhang et al., 2020).

miRNA biogenesis is precisely regulated by many protein factors at both transcriptional and post-transcriptional levels. Among them, the Mediator complex (Kim et al., 2011), CYCLIN-DEPENDENT KINASES (Hajheidari et al., 2012), CELL DEVISION CYCLE 5 (CDC5, a DNA-binding protein) (Zhang et al., 2013), Negative on TATA less 2 (NOT2) (Wang et al., 2013), SMALL1 (SMA1, a DEAD-box helicase) (Li et al., 2018b), SUPPRESSOR OF NPR1-1, CONSTITUTIVE1 (Cai et al., 2019), the PROTEIN PHOSPHATASE4 (Wang, S et al., 2019), the K homology domain protein REGULATOR OF CBF GENE EXPRESSION 3 (RCF3) (Chen et al., 2015; Karlsson et al., 2015) and the Elongator complex (Fang et al., 2015a) interact with Pol II to promote the transcription of MIRs. Notably, some of these factors such as NOT2, CDC5 and SMA1 also associate with the DCL1 complex to facilitate its co-transcriptional recruitment to pri-miRNAs. In addition, the cap-binding proteins (Kim et al., 2008), DAWDLE (DDL, a forkhead-associated domain protein) (Yu et al., 2008), NOT2, TGH, CDC5, PLEIOTROPIC REGULATORY LOCUS 1 (PRL1; an RNA binding protein) (Zhang et al., 2014), SHORT VALVE 1 (a ribosomal protein) (Li et al., 2017), the THO/TREX complex (Francisco-Mangilet et al., 2015), MODIFIER OF SNC1,2 (MOS2; an RNA-binding protein) (Wu et al., 2013), The RNA helicase MOS4-ASSOCIATED COMPLEX 7 (MAC7) (Jia et al., 2017), the ubiquitin ligase MAC3 (Li et al., 2018a), the RNA-binding protein MAC5 (Li et al., 2020) and Chromatin remodeling factor 2 (CHR2) (Wang et al., 2018) are dynamically associated with the DCL1 complex to modulate the efficient processing of pri-miRNAs. Recent studies also show that pri-miRNA stability plays a crucial role in miRNA biogenesis (Yu et al., 2008; Speth et al., 2013; Zhang et al., 2014; Fang et al., 2019; Li et al., 2020). For instance, increased pri-miRNA decay in ddl, mac3, mac5 and prl1 reduces miRNA levels (Yu et al., 2008; Zhang et al., 2014; Li et al., 2018a; Li et al., 2020). Interestingly, HYL1, besides its role in processing, is able to protect pri-miRNAs from degradation from the nuclear exosome (Gao et al., 2020). SE, another component of DCL1 complex, can mediate the decay of pri-miRNAs by nuclear localized exosome and 5’-to-3’ exoribonuclease 2 and 3 (Bajczyk et al., 2020; Gao et al., 2020; Li et al., 2020). More intriguingly, MAC5 can protect pri-miRNAs from SE-dependent degradation (Li et al., 2020). In addition, the activity of the DCL1 complex can be regulated at the transcriptional, post-transcriptional and post-translational levels. The splicing factors STABILIZED 1 (STA1) and SMA1 promote the splicing of DCL1 pre-mRNA (Ben Chaabane et al., 2013; Li et al., 2018b), while the transcription factor XAP5 CIRCADIAN TIMEKEEPER (CMA33/XCT) is required for the transcription of DCL1 (Fang et al., 2015b). HYL1 phosphorylation and dephosphorylation are also required for proper miRNA biogenesis (Manavella et al., 2012; Cho et al., 2014; Su et al., 2017; Achkar et al., 2018). Recently, light was shown to stablize DCL1 and its repressor Forhead associated domain 2 during de-etiolation in order to refine miRNA biogenesis (Choi et al., 2020; Park et al., 2021).

Squamous cell carcinoma antigen recognized by T-cells 3 (SART3) from human and its yeast counterpart precursor RNA processing, gene 24 (PRP24) are conserved spliceosome-associated proteins essential for splicing (Raghunathan & Guthrie, 1998; Bell et al., 2002; Rader & Guthrie, 2002; Liu et al., 2015; Ruegger et al., 2015). They localize at the nucleus, specifically bind U6 small nuclear RNA (U6 snRNA), and are crucial for the recycling of U4/U6 small nuclear ribonucleoproteins (Bell et al., 2002; Rader & Guthrie, 2002). In human, depletion of SART3 results in embryo lethality (Trede et al., 2007; Timani et al., 2014; Huang et al., 2018). Besides splicing, SART3 also regulates UV-induced DNA damage responses (Huang et al., 2018). Compared with human and yeast, the SART3/PRP24 homolog from plants remains to be identified and characterized.

Here we report that SE-Associated Protein 1 (SEAP1), a homolog of SART3 and PRP24 in Arabidopsis, plays important roles in miRNA biogenesis. A null seap1 mutation causes embryo lethality, whereas artificial miRNA-mediated knockdown of SEAP1 (amiR^SEAP1) results in pleiotropic developmental defects, reduced levels of miRNAs and impaired pri-miRNA splicing. SEAP1 is associated with the DCL1 complex and may contribute to the interaction of pri-miRNAs with the DCL1 complex. Notably, the levels of pri-miRNAs are also reduced in amiR^SEAP/. However, SEAP1 does not show obvious effect on pri-miRNA transcription, suggesting that it may directly or indirectly affect pri-miRNA stability. Based on these observations, we propose that SEAP1 may function as an accessory component of the DCL1 complex to facilitate miRNA biogenesis by promoting pri-miRNA splicing, processing and/or stability. We also find that intron retention of pre-messenger RNAs is increased at global levels in amiR^SEAP1, revealing that SEAP1, like SART3 and PRP24, is crucial for splicing in plants. Taken together, these results uncover both conserved and novel functions of SEAP1 in plants.

Materials and Methods

Plant Materials and Growth Condition.

The Salk_043206 (seap1-1) line is in the Columbia (Col-0) background and was obtained from the Arabidopsis Biological Resources Center (ABRC). The primers used for genotyping are listed in Supporting Information Table S1. All plants were grown at 22 °C with 16-h light and 8-h dark cycles.

Complementation Assay

A 7.5 kb genomic fragment containing the SEAP1 promoter and coding regions was PCR amplified from wild-type plants with primer pairs listed in in Supporting Information Table S1, inserted into pENTR/SD/D-TOPO (K242020, Thermo Fisher Scientific), and subsequently cloned into the binary vector pEarlyGate 303 (Earley et al., 2006) to generate pSEAP1::SEAP1-Myc. The resulting plasmid was transformed into seap1-1, and transgenic plants were identified through screening for Basta resistance.

Plasmids Construction

SEAP1 cDNA was amplified by RT-PCR, cloned into pENTR/SD/D-TOPO, and subsequently cloned into pEarlyGate203 and pMDC83 (Curtis & Grossniklaus, 2003) to generate the p35S::MYC-SEAP1 and p35S::GFP- SEAP1plasmids, respectively. The artificial miRNA was designed using WMD3 (http://wmd3.weigelworld.org/cgi-bin/webapp.cgi) (Schwab et al., 2006). A synthetic fragment containing the artificial miRNA sequence and MIR164A backbone was cloned into pENTR/SD/D-TOPO, and subsequently cloned into pMDC32. The primers used for plasmid construction are listed in Supporting Information Table S1.

Microscopy

Siliques of various developmental stages were dissected and mounted on microscope slides in a clearing agent (Visikol Inc.) overnight, and then imaged with a confocal microscope. To visualize GUS expression, samples were immersed in the GUS staining solution (1 mM 5-bromo-4-chloro-3-indoly-β-D-glucuronic acid, 100 mM NaPO₄ buffer, 3 mM each K₃Fe(CN)₆/K₄Fe(CN)₆, 10 mM EDTA and 0.1% NP-40) at 37^◦C for 5 h in the dark as described (Li et al., 2018a). The stained samples were treated with 70% ethanol to remove chlorophyll before imaging. Pollen viability was examined after Alexander’s staining (Alexander, 1969). Leaf surfaces were visualized with standard scanning electron microscopy preparative techniques.

Northern blot, RT-PCR, and Quantitative RT-PCR

Total RNAs from inflorescences were extracted with TRI reagent (Molecular Research Center). The Northern blot was performed as described (Ren et al., 2012). Five µg of total RNAs were resolved on 16% polyacrylamide gel electrophoresis (PAGE) gel and subsequently transferred to nylon membranes. The 5’ end ³²P-labeled antisense DNA oligonucleotides were used to detect miRNAs. Oligonucleotide probes used are listed in Supporting Information Table S1.

To perform RT-PCR, total RNAs were treated with DNase I (Thermo Fisher Scientific) followed by reverse transcription using M-MLV reverse transcriptase (Promega) with oligo-d(T) primers according to manufacturer’s instructions. Quantitative Real Time-PCR was performed using iCycler apparatus (Bio-Rad) with the SYBR green kit (MIDSCI). Primers used are listed in Supporting Information Table S1.

Small RNA sequencing

Small RNA libraries were prepared using total RNAs extracted from inflorescences. Two biological replicates were performed. After sequencing, miRNAs are analyzed as described (Ren et al., 2012). After removing reads aligned to t/r/sn/snoRNA, the total numbers of perfectly aligned reads were used for normalization (Nobuta et al., 2010). miRNA abundance was compared by using EdgeR with trimmed mean of M values normalization method (Robinson et al., 2010).

Differential gene expression and differential splicing analyses

RNA libraries were prepared using total RNAs extracted from inflorescences following standard protocol. Four biological replicates were performed. Raw RNA-seq reads were screened and trimmed using Trimmomatic (Bolger et al., 2014). The clean reads were mapped against to Arabidopsis reference genome (TAIR10) using STAR (Dobin et al., 2013). Intron, exon and transcript-level expression was quantified with StringTie (Pertea et al., 2015), and collected into R with Ballgown (Frazee et al., 2015). Differential gene expression analysis was conducted using R package Ballgown. Stattest function in Ballgown was used to test differential gene expression between wild types and mutants. Only the genes with standard deviation of expression larger than 1 among all samples were used in the analysis and the genes with statistic Q-value ≤ 0.05 were considered as differentially expressed genes. Differential mRNA splicing analysis was conducted using R package VaSP (Yu et al., 2021). The Single Splicing Strength (3S) score, which is junction reads normalized by gene-level average read coverage, was calculated for each annotated intron (Yu et al., 2021). Only the introns supported with at least 5 junction reads in at least 1 sample were considered, and only the genes with the average read coverage larger than 1 in all samples were used in the analysis. Student’s t-test was employed to test the difference of 3S scores between wild types and mutants. Genes with any intron at P-value ≤ 0.05 and fold change of 3S scores ≥ 2 were considered as differential splicing genes.

Detection of SE and HYL1 proteins in Arabidopsis

Protein extracts from inflorescences of Arabidopsis were separated on a 10% SDS-PAGE and detected with western blot using antibodies against HYL1 (AS06136, Agrisera,), SE (AS09532A, Agrisera) and HSC70 (ADI-SPA-818, Enzo).

BiFC assay

BiFC assay was performed as described (Klopffleisch et al., 2011). The constructs of SEAP1-cYFP, DCL1-nYFP, SE-nYFP, HYL1-nYFP and CDC5-nYFP were co-infiltrated in Nicotiana benthamiana (N. benthamiana) leaves. After 48 hours, YFP and chlorophyll autofluorescence signals were observed by a confocal microscopy (Nikon A1 HD25).

Co-IP assay

To examine the interaction of SEAP1 with DCL1 and SE, MYC-SEAP1 or SEAP1-FLAG was transiently co-expressed with DCL1-HA or MYC-SE in N. benthamiana as described (Ren et al., 2012). The expression of these transgenes was directed by a 35S promoter. Total proteins of infiltrated leaves were extracted with an extraction buffer (50 mM Tris–HCl 8.0, 150 mM NaCl, 5% glycerol, 5% Triton X-100, 1 mM EDTA, 1× complete protease inhibitor cocktail and 1 mM phenylmethylsulfonyl fluoride). Immunoprecipitation (IP) was performed on protein extracts using anti-MYC (GTA020, Bulldog Bio) or anti-FLAG antibodies (A4596, Sigma) coupled to protein G agarose beads. After IP, proteins were separated on a 10% SDS-PAGE and detected with western blot using monoclonal antibodies against MYC (M4439, Sigma), FLAG (A8592, Sigma) or HA (H6533, Sigma).

RIP Analyses.

RIP was performed as described (Ren et al., 2012). A total of ~4 g inflorescences of Col or transgenic plants harboring a p35S::MYC-SEAP1 transgene were cross-linked with 1% formaldehyde for 10 min. Then, glycine was added to quench the reaction for 10 min. Following this step, nuclei were extracted and lysed in 400 uL nuclei lysis buffer (50 mM Tris-HCl pH 8.0, 10 mM EDTA, 1% SDS) by sonication five times. After debris was removed by centrifuge at 16,000 × g for 10 min, equal amounts of proteins from various samples were diluted with RIP dilution buffer (16.7 mM Tris-HCl, 1.1% Triton X-100, 1.2 mM EDTA, pH 8.0, 167mM NaCl) and incubated with anti-MYC antibodies conjugated to protein G agarose beads or protein A/G agarose beads (for no-Ab controls). The immunoprecipitates were washed five times and then eluted with elution buffer (100 mM NaHCO3, 1% SDS) at 65 °C. Following reversing cross-linking with proteinase K (Invitrogen) and 200 mM NaCl at 65 °C, RNAs were extracted and used as templates for RT-PCR analyses, using primers listed in Supporting Information Table S1.

ChIP assay

ChIP was performed using 14-d-old seedlings from Col-0 and amiR^SEAP1 as described (Kim et al., 2011). Anti-CTD antibody (ab817, Abcam) was used for immunoprecipitation. qPCR was performed on DNAs copurified with CTD antibody, using primers listed in Supporting Information Table S1.

Results

Identification of SEAP1 as a candidate protein acting in the miRNA biogenesis

We sought to identify candidate proteins involved in miRNA biogenesis in Arabidopsis through functional gene network analyses, because functionally associated genes often act in the same biological processes (Schlitt et al., 2003). SE was used as a bait to construct a functional network using the STRING program (https://string-db.org) with a high confidence score. The SE network includes proteins known functioning in miRNA biogenesis such as DCL1, HYL1, TGH, ABH1, CPB20 BRM2, CDC5 and others, and proteins acting in RNA metabolisms (Supporting Information Fig. S1a, b). From this network, we sought to select the potential splicing-related proteins as our candidates since several splicing factors have been shown to act in miRNA biogenesis. Among several candidates, we focused on AT4G24270 because its homolog SART3 from human is functionally correlated with the PRP19 complex (Song et al., 2010), an ortholog complex of MAC, which acts in plant miRNA biogenesis. We named this protein SE-Associated Protein 1 (SEAP1). SEAP1 contains eight Half-A-Tetratricopeptide (HAT) repeats that mediates protein-protein and/or protein-RNA interactions, an RNA-binding domain and an LSM-binding motif, which is known to interact with the like-Sm (LSM) proteins (Park et al., 2016) (Fig. 1a). SEAP1 is a conserved protein in eukaryotes (Fig. 1b). To characterize SEAP1, we determined its subcellular localization by observing the green fluorescence (GF) signal in transgenic plants harboring a p35S:: GFP-SEAP1 transgene. The GFP signal was detected in the nucleus, suggesting that SEAP is a nuclear-localized protein (Fig. 1c). We also examined the expression pattern of SEAP1 in a transgenic expressing a pSEAP1::SEAP1-GUS transgene. GUS histochemical staining showed that SEAP1 was ubiquitously expressed in various tissues such as seedlings, leaves and flowers (Fig. 1d), with high levels in tissues with rapid proliferation including root and shoot tips.

Fig. 1.

SEAP1 (Serrate-Associated Protein 1) is a conserved protein essential for development. (a) A diagram of the SEAP1 protein containing eight half-a-tetratricopeptide (HAT) repeats, an RNA recognition motif (R) and an LSM (like-Sm)-binding motif. (b) Phylogenetic analysis of SEAP1 homologs in eukaryotes. The full-length protein sequences were used to construct a Neighbor-Joining tree. (c) Subcellular localization of GFP (Green Fluorescence Protein)-SEAP1 in young leaves of transgenic plants harboring p35S::GFP-SEAP1. The nuclei were visualized by DAPI staining of DNA. (d) Histochemical GUS staining of five-day-old, ten-day old seedlings and inflorescences in Col-0 (Columbia-0) harboring pSEAP1::SEAP1-GUS transgene. (e) The insertion site of seap1-1 (SALK_043206) is shown above the SEAP1 gene structure. (f) Dissected green siliques of Col and SEAP1/seap1-1. The arrows indicate aborted embryos. (g) Embryo of Col (top row) and seap1-1 (bottom row). Embryo stages: preglobular (P), globular (G), heart (H), torpedo (To), walking-stick (W), and mature embryo (M). The arrows indicate embryos arrest at the globular stage. (h) Alexander’s staining of pollens from Col-0 and seap1-1. The arrows indicate defective pollens.

SEAP1 is required for plant development

In order to determine the function of SEAP1, we obtained its transfer DNA (T-DNA) insertion mutant (SALK_043206, seap1-1), which contains a T-DNA insertion in the fourth intron of the SEAP1 gene (Fig. 1e and Supporting Information Fig. S2b). We identified the heterozygous, but not the homozygous mutants of seap1-1, from a self-pollinated population of seap1-1/+, suggesting that seap1-1 may cause embryo lethality. Indeed, ~25 % seeds were aborted in the immature siliques of heterozygous plants (Fig.1f). These aborted seeds were arrested at globular stages (Fig. 1g). Expression of a wild-type (WT) copy of SEAP1 gene under the control of its native promoter (pSEAP1::SEAP1-Myc) in seap1-1/+ fully complemented the embryo defects of seap1-1 (Supporting Information Fig. S2a, c), demonstrating that SEAP1 is essential for proper embryo development. We also determined if seap1-1 also had an effect on gametophyte transmission by analyzing the transmission rate of seap1-1 from the reciprocal crosses between seap1-1/+ and WT. When WT was used as pollen donor, seap1-1 was transmitted normally (Supporting Information Fig. S2d). However, when seap1-1 was used as pollen donor, the transmission rate of seap1-1 was ~ 46%, which was lower than expected ratio (100%) (Supporting Information Fig. S2d), suggesting that SEAP1 is crucial for male gametophyte transmission. We further evaluated the viability of pollen seap1-1/+ using Alexander’s staining (Alexander, 1969). Unlike that pollens from Col-0 were red-stained and oval-shaped, many pollens from seap1-1/+ could not be stained and appeared in irregular shapes (Fig. 1h), revealing that SEAP1 is required for pollen development.

To evaluate the post-embryonic function of SEAP1, we designed an artificial miRNA (amiR^SEAP1) to silence SEAP1 (Fig. 2a). In T1 generation, we identified amiR^SEAP1 lines, in which the transcript levels of SEAP1 were reduced (Fig. 2b). These lines displayed pleiotropic developmental defects such as smaller plant size, shorter roots and serrated leaves (Fig. 2c, d and Supporting Information Fig. S3a). Moreover, the pavement cells in amiR^SEAP1 were larger than those in Col-0 (Fig. 2e) as observed by scanning electronic microscope (SEM). To confirm the reduced SEAP1 levels were responsible for the growth defects, we expressed an amiR^SEAP1 resistant transgene of SEAP1 (SEAP-R) in amiR^SEAP1 line T5 and this transgene fully rescued the developmental defects of amiR^SEAP1 (Supporting Information Fig. S3b, c). Taken together, these results demonstrate that SEAP1 plays essential roles in the development of Arabidopsis.

Fig. 2.

SEAP1 is required for development. (a) Scheme of an artificial miRNA targeting SEAP1. Solid line represents Watson–Crick pairing and a “0” indicates a C-U mismatch pair. (b) The transcript levels of SEAP1 in amiR^SEAP1 relative to Col. The transcript levels of SEAP1 in amiR^SEAP1 lines were detected by RT-qPCR (Quantitative reverse transcription PCR), normalized to UBQ5 (Ubiquitin 5) and compared with those in Col-0 (value set as 1). Error bars: standard deviations (SD) of three replicates. **p < 0.01, *p < 0.05 (t test). (c) Twelve-day-old and twenty-five-day-old Col-0 and amiR^SEAP1. (d) Four-day-old seedlings of Col-0 and amiR^SEAP1. (e) Scanning electron microscopy of Col-0 and amiR^SEAP1 leaf surfaces.

SEAP1 promotes miRNA accumulation

Next, we examined the effect on amiR^SEAP1 on miRNA accumulation by Northern blot. Ten out of eleven examined miRNAs were reduced in abundance in amiR^SEAP1 relative to Col-0 (Fig. 3a and Supporting Information Fig. S3d). Quantitative reverse transcription PCR (RT-qPCR) analyses further confirmed the result (Supporting Information Fig. S3e), and showed that SEAP-R fully recovered miRNA levels (Supporting Information Fig. S3e). We further evaluated the impact of SEAP1 on miRNA accumulation at global levels through illumina deep sequencing. Compared with Col-0, many miRNAs were reduced in abundance in amiR^SEAP1 (Fig. 3b and Supporting Information Table S2). These results demonstrate that SEAP1 promotes miRNA accumulation.

Fig. 3.

SEAP1 is required for miRNA accumulation. (a) miRNAs levels in Col-0 and amiR^SEAP1 detected by Northern blot. The 5.8s RNA was used as a loading control. The numbers below the picture indicate the relative amounts of miRNAs in Col-0 and amiR^SEAP1. (b) Small RNA sequencing analysis in Col-0 and amiR^SEAP1. The miRNA abundance was calculated as reads per million, and a log2-transformed ratio of amiR^SEAP1/Col-0 was plotted. The box spans the first to the third quartile with the horizontal line inside the box representing the median value, and the dots represent outliers. (c) The levels of miRNA target transcripts in Col-0 and amiR^SEAP1 detected by RT-qPCR. The levels of miRNA target transcripts were normalized to those of UBQ5 and compared with Col-0 (set as 1). Error bars: standard deviations (SD) of three replicates. *p < 0.05 (Student’s t test).

We next performed RT-qPCR to examine the transcript levels of several miRNA target transcripts including SPL2, MYB33, ATHB15, PHV and TAS2, which are targets of miR156, miR159, miR166, and miR173, respectively. The levels of these targets were moderately increased in amiR^SEAP1 compared with Col-0 (Fig. 3c). This result is consistent with the decreased levels of miRNAs in amiR^SEAP1.

SEAP1 promotes the accumulation and splicing of some pri-miRNAs

Next, we asked how SEAP1 acts in miRNA accumulation. We first examined the effect of amiR^SEAP1 on pri-miRNA levels using RT-qPCR analyses. The result showed that eight out of nine examined pri-miRNAs were reduced abundance relative to WT (Fig. 4a). Because amiR^SEAP1 is a weak allele of SEAP1, this result suggests that SEAP1 likely promotes pri-mRNA accumulation. Since pri-miRNA levels are partially determined by transcription, we next asked if SEAP1 affected pri-miRNA transcription by examining the occupancy of Pol II at the MIR promoters with a chromatin immunoprecipitation (ChIP) assay followed by qPCR using antibodies against CTD of the largest subunit of Pol II. The result showed that amiR^SEAP1 did not affect the occupancy of Pol II at the promoters of pri-miRNAs (Fig. 4b). Moreover, we did not observe the interaction of SEAP1 with Pol II (Fig. 4c), suggesting that SEAP1 may not influence pri-miRNA transcription. Next, we explored if amiR^SEAP1 could reduce pri-miRNA stability, which is another factor contributing to pri-miRNA levels. To test this possibility, we examined the half-lives of pri-miRNAs using cordycepin as a transcription inhibitor in Col-0 and amiR^SEAP1. The half-lives of four examined pri-miRNAs in amiR^SEAP1 were similar to/ slightly shorter than those in Col-0 (Supporting Information Fig. S4a–d). Because pri-miRNA processing efficiency is likely lower in amiR^SEAP1 (see below), which shall increase pri-miRNA half-lives, this result suggests that pri-miRNA stability is reduced in amiR^SEAP1.

Fig. 4.

SEAP1 is required for the accumulation and splicing of some pri-miRNAs. (a) The accumulation of pri-miRNAs in Col-0 and amiR^SEAP1 detected by RT-qPCR. Pri-miRNA levels in amiR^SEAP1 were normalized to those of UBQ5 and compared with Col-0 (set as 1). Error bars: standard deviations (SD) of three replicates. *p < 0.05 (Student’s t test). (b) The occupancy of Pol II (Ploymerase II) at MIR promoters in Col-0 and amiR^SEAP1 detected by chromatin immunoprecipitation (ChIP) followed by qPCR. IP was performed using antibodies against CTD (C-terminal domain) on protein extracts from Col-0 and amiR^SEAP1. The intergenic region between At2g17470 and At2g17460 (POL II C1) was amplified as a negative control. (c) The interaction between SEAP1 and RPB1. MYC-SEAP1 was IPed from inflorescence protein extracts of transgenic plants harboring MYC-SEAP1 using anti-MYC antibodies. MYC-SEAP1 and RPB1 were detected by western blot using antibodies recognizing MYC and RPB1, respectively. (d) Diagram showing the partial regions of several pri-miRNAs examined in (e). Arrows indicate primers used for PCR. The numbers in red indicate the length of PCR products without intron splicing. The numbers in black show the intron length. (e) Intron-retention of pri-miRNAs analyzed by RT-PCR. Upper arrows indicate the unspliced transcripts. Lower arrows indicate the intron-spliced transcripts. ACT2 (Actin 2) served as the loading control. G: genomic DNA. The numbers shown below the figures indicate the ratio of the levels of spliced transcripts to those of total transcripts.

Because SART3 and PRP24 act in splicing, we next examined the effect of amiR^SEAP1 on the splicing of intron-containing pri-miRNAs by RT-PCR using primer pairs spanning the intron region of pri-miRNAs (Fig. 4d). Among four examined pri-miRNAs, two had increased intron retention in amiR^SEAP1 compared with Col-0 (Fig. 4e), suggesting that the splicing of some pri-miRNAs is impaired in amiR^SEAP1. The splicing defect of MIR171b was further confirmed by RNA-seq analysis (Supporting Information Fig. S4e).

SEAP1 interacts with SE

To further investigate how SEAP1 affects miRNA biogenesis, we examined the effect of amiR^SEAP1 on the expression and splicing patterns of DCL1, HYL1, SE, DDL1 and HEN1, which act in miRNA biogenesis. The transcript levels of SE and HEN1 were slightly increased, and those of DCL1, HYL1 and DDL1 were not significantly altered in amiR^SEAP1 (Supporting Information Fig. S5a). The splicing patterns of these genes were not affected in amiR^SEAP1 (Supporting Information Fig. S5b, c). Western blot analyses further showed that the protein levels of HYL1 and SE were not changed in amiR^SEAP1 relative to WT (Supporting Information Fig. S5d).

Some splicing-related proteins have been shown to be associated with the DCL1 complex. Thus, we performed Bimolecular fluorescence complementation (BIFC) assay to test the potential interaction of SEAP1 with DCL1, HYL1, SE and CDC5. SEAP1 was fused with the N-terminal half of YFP (SEAP1-nYFP), and DCL1, HYL1, SE and CDC5 with the C-terminal half of YFP (DCL1-cYFP, HYL1-cYFP, SE-cYFP and CDC5-cYFP). Co-expression of SEAP1-nYFP with HYL1-cYFP, SE-cYFP or CDC5-nYFP produced YFP signals (Fig. 5a). The results indicate that SEAP1 might interact with HYL1, SE and CDC5, but not DCL1. Next, we performed co-IP assays to confirm the interaction of SEAP1 with DCL1, HYL1, CDC5 and SE. We transiently co-expressed MYC-SEAP1 with DCL1-HA, SEAP1-FLAG with MYC-SE and SEAP1-GFP with MYC-CDC5 or HYL1-FLAG in N. benthamiana and performed IP with anti-MYC, -FLAG or -GFP antibodies. After IP, SE and CDC5 were detected in the SEAP1-FLAG and SEAP1-GFP precipitates respectively whereas HYL1 and DCL1 were not in MYC-SEAP1 and SEAP1-GFP precipitates respectively (Fig. 5b,c and Supporting Information Fig. S6a, b). Although the Bi-FC signal between SEAP1 and HYL1 was observed, we did not detect their co-IP, revealing that SEAP1 does not directly interact with HYL1. It may be weakly and indirectly associated with HYL1. To further validate the result, we co-expressed SEAP1-MYC with DCL1-HA, CDC5-HA or SE-HA in Arabidopsis mesophyll protoplasts and performed IP with anti-MYC antibodies. SEAP1 co-IPed with SE and CDC5 but not DCL1 (Supporting Information Fig. S6c,d). We also examined the SE-SEAP1 interaction using recombined MBP (Maltose-binding protein)-SEAP1 and 6XHIS (Histidine)-SE. The result showed that MBP-SEAP1, but not MBP, was able to pull down SE (Supporting Information Fig. S6e). These results show that SEAP1 interacts with SE and CDC5.

Fig. 5.

SEAP1 associates with DCL1 (Dicer-like 1) complex and affects the interaction between HYL1 and pri-miRNAs. (a) Bimolecular fluorescence complementation (BiFC) analysis of SEAP1 with DCL1, HYL1, SE, and CDC5. Paired cYFP- and nYFP-fusion proteins were co-expressed in tobacco leaves. Yellow color indicates the BiFC signal detected by a confocal microscopy at 48 h after infiltration. (b) Co-IP between SEAP1 and SE. SEAP1-FLAG or GFP-FLAG was co-expressed with MYC-SE in tobacco leaves. IPs were performed using anti-FLAG antibodies. SEAP1-FLAG, MYC-SE and GFP-FLAG were detected by western blot. (c) Co-IP between SEAP1-GFP and MYC-CDC5. SEAP1-GFP or GFP was co-expressed with MYC-CDC5 in tobacco leaves. IPs were performed using anti-GFP antibodies. SEAP1-GFP, GFP and MYC-CDC5 were detected by western blot. (d) The association of HYL1 in amiR^SEAP1 relative to Col detected by RNA immunoprecipitation. IP was performed with the anti-HYL1 antibodies. Detection of HYL1 after IP was shown in upper panel. Ten percent of IPs and 2% input proteins were detected by Western blot. Pri-miRNAs associated with HYL1 were examined by RT-qPCR and normalized to the input. UBQ5 serves as a negative control. *P < 0.05 (Student’s t test).

The interaction of HYL1 with some pri-miRNAs is impaired in amiR^SEAP1

The interaction of SEAP1 with SE prompted us to test if amiR^SEAP1 affect the interaction of HYL1 with pri-miRNAs using an RNA IP (RIP) assay, which has been used as indication of pri-miRNA loading to the DCL1 complex (Ren et al., 2012). We IPed HYL1 from protein extracts of amiR^SEAP1 and WT using antibodies recognizing HYL1 (Fig. 5d), and performed RT-qPCR to examine the amounts of pri-miRNAs associated with HYL1. The result showed that four out of six examined HYL1-bound pri-miRNAs were reduced in abundance in amiR^SEAP1 relative to WT (Fig. 5d). Given the fact that amiR^SEAP1 is a weak allele, this result shows that SEAP1 may facilitate pri-miRNA loading to the DCL1 complex, and thereby promote pri-miRNA processing.

SEAP1 affects gene expression at global levels

We also asked if SEAP1 has a general impact on gene expression. We compared mRNA profiling in the flowers from the amiR^SEAP1 with that in WT by RNA-seq analyses. Four biological replicates were performed. After sequencing, we identified differentially expressed genes (DEGs) between amiR^SEAP1 lines and Col-0 with fold change of 1.5 or more. A total of 313 up-regulated and 451 down-regulated genes were identified, respectively, in amiR^SEAP1 (Fig. 6a and Supporting Information Table S3). To better understand the function of SEAP1, we performed gene ontology (GO) analyses on DEGs in amiR^SEAP1. Genes related to the regulation of enzymatic activity, polysaccharide catabolism, organelle organization, cell wall modification, pollen tube growth, and RNA processing were enriched in down-regulated genes (Fig. 6b and Supporting Information Table S4), while genes involved in indoleacetic acid (IAA) biogenesis, glucosinolate metabolism, response to stress and RNA destabilization were detected in up-regulated genes (Fig. 6c and Supporting Information Table S4).

Fig. 6.

SEAP1 affect gene expression at global levels. (a) Venn diagram showing the degree of overlap of splicing defective genes with up-regulated or down-regulated in amiR^SEAP1 relative to Col-0. (b) Gene ontology (GO) enrichment of down-regulated genes in amiR^SEAP1 relative to Col-0. (c) GO enrichment of up-regulated genes in amiR^SEAP1 relative to Col-0.

SEAP1 promotes pre-mRNA splicing

The effect of SEAP1 on pri-miRNA splicing suggests that like its homologs in yeast and human, SEAP1 may have a role in pre-mRNA splicing. Since intron retention is a major splicing defection, we examined the effect of amiR^SEAP1 on intron retention of pre-mRNAs. The ratio between RNA-seq reads mapped to introns and those mapped to exons was used to calculate the intron retention. We measured the intron retention at global levels using all annotated transcripts passed the abundance filter and found that the intron retention rate in amiR^SEAP1 was higher than in WT (Fig. 7a). A total of 343 genes were found to have higher intron retention rate in amiR^SEAP1 relative to WT (Fig. 6a, 7a and Supporting Information Table S5). Fig. 7b showed two examples of impaired mRNA intron retention in amiR^SEAP1 relative to WT. We randomly selected 6 genes with increased intron retention in amiR^SEAP1 for validation using RT-PCR analysis (Fig. 7c). The intron retention was increased in all these six genes in amiR^SEAP1 (Fig. 7d), agreeing with the RNA-seq result. We next asked if splicing defects caused altered expression levels in amiR^SEAP1. The co-occurrence of genes with intron retention defects with both up- and down regulated DEGs were examined. Only a very small portion of DEGs has splicing defects (Fig. 6a).

Fig. 7.

SEAP1 affects splicing at global levels. (a) Genome–wide intron Single Splicing Strength (3S, see Methods) scores in Col-0 and amiR^SEAP1. The red dots indicate introns with significant differential splicing between Col-0 and mutants and the gray dots indicate no significant difference. (b) Two examples of differentially spliced transcripts between Col-0 and amiR^SEAP1. For each transcript, the x-axis is the genomic position, the y-axes are the RNA-seq read numbers, the arcs indicate exon-exon junctions (introns) and the numbers are 3S scores. Transcripts from Col are in red. Transcripts from amiR^SEAP1 are in pink and gene structures are in blue. The significant differentially spliced introns are highlighted in vertical gray box. (c) Diagram showing the structures of several differentially spliced transcripts selected for validation. Arrows indicate primers used for RT-PCR. Arrows indicated primer positions used for PCR in (d). The numbers in red indicate the length of PCR products without intron splicing. The numbers in black show the intron length. (d) Intron-retention analysis of selected transcripts by RT-PCR using primers pairs indicated in (c). ACT2 used as the loading control. G: genomic DNA. The numbers indicate the ratio of the levels of spliced transcripts to those of total transcripts.

SEAP1 directly binds to U6 and affects the abundance of U6

In human, SART3 binds U6 snRNA and is required for U6 snRNA accumulation (Liu et al., 2015; Ruegger et al., 2015). We thus asked if SEAP1 has a similar function in splicing. We examined the interaction of SEAP1 with U6 snRNA by RIP on the plants harboring the MYC-SEAP1 transgene using anti-MYC antibodies. RT-PCR detected the presence of U6 snRNA but not U4 snRNAs in MYC-IPs, demonstrating that SEAP1 indeed binds U6 snRNAs (Supporting Information Fig. S7a). We also asked if it could bind pri-miRNAs in vivo by examining the enrichment of pri-miRNAs in SEAP1 IP. Among four examined pri-miRNAs, two were detected in the SEAP1 IP but not in the no-antibody controls (Supporting Information Fig. S7b), which suggests that SEAP1 associates with some pri-miRNAs in vivo.

We next used northern blot to examine the effect of amiR^SEAP1 on U6 snRNAs. The abundance of U6 was reduced in amiR^SEAP1 lines relative to Col-0 while that of U4 was slightly increased (Supporting Information Fig.S7c). These results suggested that SEAP1 is likely an ortholog of SART3 in splicing.

Discussion

SEAP1 is a conserved protein in eukaryotes. Its orthologs, SART3 and PRP24 from metazoans and yeast are associated with the U4 and U4–U6 di-snRNP and are required for the recycling of U4 and U6 components from tri-snRNP (Raghunathan & Guthrie, 1998; Bell et al., 2002; Rader & Guthrie, 2002; Liu et al., 2015; Ruegger et al., 2015). However, its function in plants remains to be identified. Here we show that a T-DNA insertion in SEAP1 causes embryo lethality, and downregulation of SEAP1 in amiR^SEAP1 causes pleiotropic developmental defects. These results demonstrate that SEAP1 is essential for plant development. We find that the intron retention is increased in amiR^SEAP1 at global levels, showing that like its orthologs, SEAP1 functions in pre-mRNA splicing (Fig. 8). Interestingly, functional network analysis shows that SEAP1 may be associated with SE and may function in miRNA biogenesis. In fact, amiR^SEAP1 reduces the accumulation of miRNAs. Moreover, SEAP1 interacts with SE and CDC5, revealing that SEAP1 may act as their co-factors or accessory factors to function in miRNA biogenesis (Fig. 8). We also observed that the levels of several examined miRNA target transcripts are increased. Taken together, these results suggest that SEAP1 may partially regulate development through promoting miRNA biogenesis and splicing.

Fig. 8.

A proposed model for the roles of SEAP1. SEAP1 acts in both pre-miRNA splicing and miRNA biogenesis. In miRNA biogenesis, SEAP1 may modulate the splicing of some pri-miRNAs, facilitate pri-miRNA processing as an accessory factor of the DCL1 complex, and indirectly or directly stabilize pri-miRNAs.

SEAP1 plays at least three roles in miRNA biogenesis. SEAP1 may modulate miRNA biogenesis through its interaction with the DCL1 complex. BiFC analyses show the possible interaction of SE, CDC5 and HYL1 with SEAP1. Co-IP assays confirmed the interaction of SE and CDC5, but not HYL1, with SEAP1, indicating that SEAP1 may weakly or dynamically associate with the DCL1 complex. Several splicing related proteins including ILP1, NTR1, MAC7, MAC3, CDC5, PRL1, SMA1, and others, interact with the DCL1 complex to facilitate miRNA biogenesis by promoting the assembly and/or activities of the DCL1 complex (Zhang et al., 2013; Zhang et al., 2014; Jia et al., 2017; Li et al., 2018a; Li et al., 2018b; Wang, J et al., 2019a). By analogy, we believe that SEAP1 may also act as an accessory component of the DCL1. In fact, the interaction of HYL1 and pri-miRNAs is reduced in amiR^SEAP1, revealing that SEAP1 may promote the efficient recognition of pri-miRNAs by the DCL1 complex. Since amiR^SEAP1 is a weak allele, SEAP1 likely have more profound impacts on pri-miRNA loading than observed. In addition, SEAP1 may affect miRNA accumulation by promoting pri-miRNA accumulation. In amiR^SEAP1, pri-miRNA levels are reduced. This could be caused by reduced pri-miRNA transcription or stability. However, SEAP1 does not affect the occupancy of Pol II at MIR promoters and does not interact with Pol II, arguing against a potential role of SEAP1 in regulating MIR transcription. The half-lives of pri-miRNAs are similar or slightly lower in amiR^SEAP1 than in Col. This result reveals that pri-miRNA decay may be increased in amiR^SEAP1, since a less efficient HYL1-pri-miRNA interaction reduces pri-miRNA processing efficiency, which shall increase pri-miRNA abundance and slow pri-miRNA decay. Thus, we prefer that SEAP1 may indirectly or directly enhance pri-miRNA stability (Fig. 8), although we cannot completely rule out the potential impact of SEAP1 on transcription elongation of pri-miRNAs. Supporting this notion, the transcript levels of some genes involved in RNA destabilization are increased (Fig. 6c). It has been documented that splicing is an essential factor for the processing of some pri-miRNAs (Stepien et al., 2017). We find that intron retention of some pri-miRNAs is altered in amiR^SEAP1. Thus, it is possible that SEAP1 may also affect the processing of some pri-miRNAs by modulating their splicing.

SEAP1 is an RNA binding protein. Like SART3 and PRP24, SEAP1 interacts with U6 snRNA. Interestingly, RIP assay also shows that SEAP1 binds two out of four examined pri-miRNAs, suggesting that SEAP1 interacts with a subset of pri-miRNAs. In yeast, PRP24 binds U6 snRNA to unwind its internal stem loop in an adenosine triphosphate-independent manner, which is required for U6 entering into the slicing cycle (Raghunathan & Guthrie, 1998; Martin-Tumasz et al., 2011). Therefore, we reason that SEAP1 may recognize some pri-miRNAs through its RNA-binding domains, and unwind undesired structures of these pri-miRNAs to facilitate their processing. Supporting this notion, SART3 has been found to specifically interact with pre-miR34a and promotes its efficient processing in human (Sherman et al., 2019).

Our RNA-seq analyses identified intron retention defects of many pre-mRNAs in amiR^SEAP1, revealing that SEAP1 acts in splicing. SEAP1 binds U6, but not U4 snRNA, and is required for U6 RNA accumulation, which resembles the observations in yeast and human (Blanton et al., 1992; Liu et al., 2015; Ruegger et al., 2015). Thus, SEAP1 likely acts similarly to SART3 and PRP24 in splicing. RNA-seq analysis also revealed that many genes are altered expression in amiR^SEAP1. Interestingly, GO analysis suggests that some genes involved in pollen development are down-regulated in amiR^SEAP1 (Fig. 6b), agreeing with the role of SEAP1 in pollen development. No obvious correlations between intron retention and alterations in gene expression were detected. The expression levels of genes with intron retention were elevated, decreased or unchanged in amiR^SEAP1. Therefore, SEAP1 may have additional roles in regulating gene expression besides modulating splicing and miRNA biogenesis.

In conclusion, our work not only shows that SEAP1, like SART3 and PRP24, is a key splicing factor, but also identifies SEAP1 as an important component in miRNA biogenesis. It should be noted that not all miRNAs/pri-miRNAs and mRNA slicing in the amiR^SEAP1 lines are affected. A likely scenario is that the amiR^SEAP1 used in this study is a weak allele, in which the SEAP1 transcript level is ~ 40% of WT, given the facts that SEAP1 is universally expressed, interacts with SE in miRNA biogenesis and U6 in splicing. Alternatively, SEAP1 may need co-factors that function in a locus specific manner. In miRNA biogenesis, SEPA1 may act as an accessory factor to enhance pri-miRNA processing. In addition, it may also promote pri-miRNA accumulation by modulating pri-miRNA stability and/or splicing. Besides SEAP1, multiple splicing factors acting in different steps of splicing also associate with the DCL1 complex. Some of them also interact with Pol II, while the others do not. These splicing factors affects pri-miRNA transcription, splicing, stability, and/or processing, suggesting that various steps of miRNA biogenesis may be coordinately regulated by splicing factors to ensure proper miRNA production. We speculate that these factors may be recruited to pri-miRNAs at various stages of pri-miRNA transcription, which enables these factors to transiently or dynamically interact with the DCL1 complex, facilitating co-transcriptionally pri-miRNA processing.

Data Availability

All Sequencing data were submitted to National Center for Biotechnology Information Gene Expression Omnibus. Accession numbers for small RNAs are: GSM5235921 and GSM5235922 for Col, and GSM5235923 and GSM5235924 for amiR^SEAP1. Accession numbers for transcriptome are: GSM5235913, GSM5235914, GSM5235915 and GSM5235916 for Col, and GSM5235917, GSM5235918, GSM5235919 and GSM5235920 for amiR^SEAP1.