Posttranslational modification (PTM) of intracellular proteins by O-linked N-acetylglucosamine (O-GlcNAc) is an important regulatory PTM that modulates protein activities and thereby controls cellular functions according to nutrient and energy status1,2. Extensive studies in animals have shown that thousands of proteins involved in diverse biological processes are modified on serine and threonine residues by O-GlcNAcylation, which is catalyzed by O-GlcNAc transferase (OGT) using UDP-GlcNAc as donor substrate1,3. As a sensor of primary metabolic status, O-GlcNAcylation plays key roles in cellular homeostasis and responses to nutritional and stress factors1,2,4,5, whereas dysregulation of O-GlcNAcylation has been implicated in many diseases including cancer, diabetes, cardiovascular and neurodegenerative diseases5,6. The Arabidopsis genome encodes two OGT homologs: SPINDLY (SPY) and SECRET AGENT (SEC). The spy mutant was identified as a gibberellin (GA) response mutant with phenotypes of enhanced seed germination, early flowering, increased stem elongation, and hyposensitivity to the stress hormone abscisic acid (ABA)7,8. The sec mutants show no dramatic phenotype, but the double loss-of-function spy sec mutants are embryo lethal9. SEC and SPY were recently reported to have O-GlcNAc and O-fucosyl transferase activities, respectively, and they antagonistically regulate DELLAs, the repressors of GA signaling10. The lethal phenotype of spy sec double mutants suggests that SPY and SEC have broader functions, which remain to be investigated at the molecular level10,11,12. Our recent study identified the first large set of 971 O-GlcNAcylated peptides in 262 Arabidopsis proteins13. The functions of these O-GlcNAcylation events remain to be characterized.

One of the O-GlcNAcylated proteins is AtACINUS, an Arabidopsis homolog of the mammalian apoptotic chromatin condensation inducer in the nucleus (Acinus)14. In animals, Acinus forms the apoptosis and splicing-associated protein (ASAP) complex by recruiting RNA-binding protein S1 (RNPS1), a peripheral splicing factor, and Sin3-associated protein of 18 kDa (SAP18), a chromatin remodeler, through its conserved RNPS1–SAP18 binding (RSB) domain14. Another RSB-containing protein, Pinin, forms a similar protein complex named PSAP, which has distinct biological functions14,15. The ASAP and PSAP complexes are believed to function at the interface between histone modification, transcription, and alternative splicing (AS) in metazoans14,16,17. In Arabidopsis, AtRNPS1, also known as ARGININE/SERINE-RICH 45 (SR45), has been implicated in splicing, transcription, and RNA-dependent DNA methylation, with effects on multiple aspects of plant development as well as stress and immune responses18,19,20,21,22,23. AtSAP18 has been shown to associate with transcription factors involved in stress responses and embryo development24,25. AtACINUS, AtSAP18, and SR45 have been shown to associate with a transcription factor involved in flowering26. While sequence analysis predicted similar ASAP complex in plants23, interactions among SR45, AtSAP18, and AtACINUS remain to be tested experimentally and the functions of AtACINUS and AtPININ remain to be characterized genetically.

Our finding of O-GlcNAcylation of AtACINUS suggests that the functions of AtACINUS are regulated by O-linked glycosylation13. We therefore performed genetic, genomic, and proteomic experiments to understand the functions of AtACINUS and its regulation by O-linked sugar modifications. Our results demonstrate key functions of AtACINUS and its distant homolog AtPININ in regulating seed germination, ABA sensitivity, and flowering, through direct involvement in AS of two key components of the ABA signaling pathway and in the transcriptional regulation of the floral repressor FLC. Our results further show that AtACINUS is modified by both O-GlcNAc and O-fucose, is part of the ASAP complex, and associates with splicing and transcription factors. A subset of AtACINUS-dependent AS events is altered in the spy and sec mutants, providing genetic evidence for regulation of AS by the O-linked glycosylations.


AtACINUS and AtPININ play genetically redundant roles

The Arabidopsis AtACINUS (AT4G39680) protein is 633 amino-acid long, and it shares sequence similarity to all the known motifs of the human Acinus including the N-terminal SAF-A/B, Acinus, and PIAS (SAP) motif, the RNA-recognition motif (RRM), and the C-terminal RSB motif (Fig. 1a and Supplementary Fig. 1a)14,16,27. AtACINUS is a unique gene in Arabidopsis with no homolog detectable using standard BLAST (Basic Local Alignment Search Tool) search of the Arabidopsis protein database. However, another Arabidopsis gene (AT1G15200, AtPININ) encodes a protein with an RSB domain and is considered a homolog of mammalian Pinin14. AtACINUS and AtPININ share 12 amino acids within the 15-amino acid region of the RSB motif (Fig. 1b), but no sequence similarity outside this motif.

Fig. 1: AtACINUS and AtPININ are genetically redundant.
figure 1

a Diagrams of the domain structures of AtACINUS and AtPININ. SAP: SAF-A/B, Acinus, and PIAS motif. RRM: RNA-recognition motif. RSB: RNPS1–SAP18 binding domain. G and F indicates the position of O-GlcNAcylation and O-fucosylation modifications, respectively. b The sequence alignment of the RSB domains of AtACINUS and AtPININ. Conserved amino acids are highlighted in green. c Diagrams of the AtACINUS and AtPININ (translation start at position 1) with T-DNA insertion sites in acinus-1, acinus-2, and pinin-1 mutants. d Plant morphologies of wild-type (WT), acinus-1, acinus-2, pinin-1, acinus-1 pinin-1, and acinus-2 pinin-1 grown on soil for 20 days. e Five-week-old WT, acinus-1, acinus-2, pinin-1, acinus-1 pinin-1, and acinus-2 pinin-1 plants grown under long day condition. Inset shows enlarged view of the acinus-1 pinin-1 and acinus-2 pinin-1 mutants. f Expression of either AtACINUS-GFP or YFP-AtPININ suppresses the growth defects in acinus-2 pinin-1 double mutant (ap).

To study the biological function of AtACINUS, we obtained two mutant lines that contain T-DNA insertions in the exons of AtACINUS, Salk_078554 and WiscDsLoxHs108_01G, which are designated acinus-1 and acinus-2, respectively (Fig. 1c). These mutants showed no obvious morphological phenotypes except slightly delayed flowering (Fig. 1d, e). The weak phenotype of acinus is surprising considering the important function of its mammalian counterpart and the absence of any close homolog in Arabidopsis.

We did not expect AtACINUS and AtPININ to have redundant functions, considering their very limited sequence similarity and the fact that mammalian Acinus and Pinin have distinct functions14. AtPININ shares extensive sequence similarity with human Pinin surrounding the RSB domain14 (Supplementary Fig. 1b). Phylogenetic analysis indicated that AtPININ and human Pinin belong to one phylogenetic branch that is distinct from that of AtACINUS and human Acinus (Supplementary Fig. 1c), suggesting independent evolution of ACINUS and PININ before the separation of the metazoan and plant kingdoms. However, Pinin can, through its RSB domain, interact with RNPS1 and SAP18 to form a complex (PSAP) similar to the ASAP complex. Therefore, we tested the possibility that the weak phenotype of Arabidopsis acinus mutants is due to functional redundancy with AtPININ.

We obtained a T-DNA insertion mutant of AtPININ (pinin-1, T-DNA line GABI_029C11). The pinin-1 mutant also showed no obvious morphological phenotype (Fig. 1d). We then crossed pinin-1 with acinus-1 and acinus-2 to obtain double mutants. Both acinus-1 pinin-1 and acinus-2 pinin-1 double mutants displayed pleiotropic phenotypes including severe dwarfism, short root, pale leaves, narrow and twisted rosette leaves with a serrated margin, severely delayed flowering, altered phyllotaxis, increased numbers of cotyledons and petals, and reduced fertility (Fig. 1d, e and Supplementary Fig. 2). The acinus-2 pinin-1 double mutants transformed with 35S::AtACINUS-GFP or 35S::YFP-AtPININ displayed near wild-type (WT) morphology (Fig. 1f), confirming that the phenotypes of the double mutants are due to loss of both AtACINUS and AtPININ, and the two genes play genetically redundant roles. The AtACINUS-GFP and YFP-AtPININ proteins are localized in the nucleus outside the nucleolus (Supplementary Fig. 3).

We also noticed that the seed germination was delayed in the acinus pinin mutant (Fig. 2a). This, together with the pale leaf and dwarfism phenotypes, suggests an alteration in ABA response. Indeed, on 0.25 μmol/L ABA, germination of the acinus-2 pinin-1 double mutant seeds was further delayed compared to the WT and the single mutants (Fig. 2b). Dose response experiment indicates that seed germination of the acinus-1 pinin-1 and acinus-2 pinin-1 double mutants is about threefold more sensitive to ABA than WT and the acinus and pinin single mutants (Fig. 2c). Similarly, post-germination seedling growth of acinus-2 pinin-1 was more inhibited by ABA (Supplementary Fig. 4a). These ABA-hypersensitive phenotypes were rescued by expression of either AtACINUS-GFP or YFP-AtPININ in the acinus-2 pinin-1 background (Fig. 2d and Supplementary Fig. 4b). These results indicate that the acinus-2 pinin-1 double mutant is hypersensitive to ABA, and that AtACINUS and AtPININ are redundant negative regulators of ABA responses.

Fig. 2: The acinus pinin double mutants showed ABA hypersensitive phenotypes.
figure 2

a, b Germination rates of WT, acinus-2, pinin-1, and acinus-2 pinin-1 after different days on ½ MS medium without ABA (a) or with 0.25 μmol/L ABA (b). The data points of WT, acinus-2, and pinin-1 overlap. Statistically significant differences between WT and acinus-2 pinin-1 were determined by two-tailed t test. The P values for a and b in a are 1.24E−5 and 2.96E−2. The P values for a, b, c, d, e, f, g, h, and i in b are 7.69E−3, 5.60E−6, 2.88E−6, 3.92E−4, 3.17E−4, 6.00E−3, 3.10E−2, 4.77E−2, and 8.57E−3. c Seed germination rates of the indicated genotypes on ½ MS medium supplemented with increasing concentrations of ABA after 5 days. Note that the data points of acinus-1 pinin-1 and acinus-2 pinin-1 overlap and those of WT, acinus-1, acinus-2, and pinin-1 overlap. Statistically significant differences between WT and acinus-2 pinin-1 were determined by two-tailed t test. The P values for a, b, and c are 3.51E−2, 5.14E−8, and 2.99E−2. d Seed germination and development of the indicated genotypes on ½ MS medium with or without 0.5 μmol/L ABA. The pictures were taken 6 days after germination. Values represent mean ± SD calculated from three biological replicates (n = 3) for ac.

AtACINUS and AtPININ are involved in AS of specific introns

We conducted an RNA-seq analysis of the transcriptome of the acinus-2 pinin-1 double mutant. WT and acinus-2 pinin-1 seedlings were grown under constant light for 14 days, and RNA-seq was performed with three biological replicates, each yielding a minimum of 22.4 million uniquely mapped reads. The RNA-seq data confirmed the truncation of the AtACINUS and AtPININ transcripts in the double mutant (Supplementary Fig. 5). Compared to WT, the acinus-2 pinin-1 double mutant showed significantly decreased expression levels for 786 genes and increased levels of 767 genes (fold change >2, multiple-testing corrected p-value <0.05), which include the flowering repressor FLC28 (Supplementary Data 1).

A significantly higher proportion of reads was mapped to the intron regions in the acinus-2 pinin-1 double mutant than in the WT (Supplementary Fig. 6a). Further analyses using the RACKJ software package revealed an increase of retention of 258 introns in 225 genes and decreased retention of 31 introns in 31 genes in the acinus-2 pinin-1 double mutant compared to WT (Fig. 3a and Supplementary Data 2). Intron retention was the dominant form of splicing defect in the acinus-2 pinin-1 double mutant (Fig. 3a and Supplementary Fig. 6b). About 99% of these genes contain multiple introns, and the defects tend to be retention of a specific single intron among many introns of each gene, indicating defects in AS rather than general splicing. Among the RNAs showing increased intron retention, 26 RNAs also showed decreased levels of RNA abundance, and their retained introns introduce in-frame stop codons (Supplementary Fig. 7), consistent with non-sense-mediated decay29. The results show that AtACINUS and AtPININ function in AS, primarily by enhancing splicing of a specific intron among many introns of each transcript.

Fig. 3: RNA-sequencing analysis of acinus-2 pinin-1 showed differential intron retention and expression level of many genes.
figure 3

a Number of introns that showed increased or decreased intron retention in acinus-2 pinin-1 and the number of genes that contain these introns. b Comparison between genes differentially expressed in acinus-2 pinin-1 and ABA-responsive genes. RNA-seq was conducted using 14-day-old light-grown seedlings for both genotypes.

We found a significant overlap between ABA-induced genes and the genes overexpressed in acinus-2 pinin-1 (p-value by random chance <2.42E−13) (Fig. 3b). Only four of these RNAs were mis-spliced in acinus-2 pinin-1. One possibility is that intron retention in RNAs encoding components of ABA synthesis or signaling pathway leads to expression of ABA-responsive genes. Indeed, we found retention of the 10th intron of ABA HYPERSENSIVE 1 (ABH1) in the acinus-2 pinin-1 double mutant (Fig. 4a).

Fig. 4: ABH1 and HAB1 showed increased intron retention in acinus-2 pinin-1 and ABH1 and HAB1 mRNAs are associated with AtACINUS.
figure 4

a Integrative genomic viewer (IGV) display of increased intron retention (IR) of the ABH1 10th intron in acinus-2 pinin-1 compared to WT. b RT-PCR of ABH1 in 12-day-old seedlings of the indicated genotypes using primers at positions indicated by arrowheads in a. c Intron retention ratio of ABH1 10th intron as determined by RT-qPCR in 12-day-old seedlings of the indicated genotypes. The intron-containing form ABH1.2 was highly accumulated while the spliced form ABH1.1 was reduced in acinus-2 pinin-1 compared to WT, the single mutants, or the double mutant complemented by YFP-AtPININ or AtACINUS-GFP. Statistically significant differences to WT or between indicated genotypes were determined by two-tailed t test. The P values for a, b, c, d, e, f, and g are 1.97E−2, 5.02E−3, 1.94E−3, 4.90E−4, 9.01E−4, 1.39E−3, and 2.32E−3. d RT-PCR of HAB1 in 12-day-old WT and acinus-2 pinin-1 seedlings treated with ABA (100 μmol/L for 3 h). e RT-qPCR quantification of the fold changes of expression levels of each splice forms of HAB1 after ABA treatment of 12-day-old WT and acinus-2 pinin-1 seedlings. The P values for a, b, and c are 2.29E−2, 3.08E−3, and 1.60E−3. f, g Quantification of ABH1 and HAB1 mRNAs by qPCR after RNA-IP using α-GFP antibody in 7-day-old AtACINUS-GFP/acinus-2 seedlings, compared to 35S::GFP as a negative control. Statistical significance was determined by two-tailed t test. The P values for a and b in f are 2.24E−2 and 2.31E−2. The P values for a and b in g are 2.18E−2 and 3.54E−2. Values represent mean ± SD calculated from three biological replicates (n = 3) for c, e, f.

ABH1 encodes the large subunit of the dimeric Arabidopsis mRNA cap-binding complex (NUCLEAR CAP-BINDING PROTEIN SUBUNIT 1, CBP80) and functions as a negative regulator of ABA responses including inhibition of seed germination30,31. The retention of the 10th intron of ABH1 introduces a premature stop codon that truncates the C-terminal 522 amino acids of ABH1 (Fig. 4a). Quantification using qRT-PCR analysis in 12-day-old seedlings showed that the intron-containing ABH1.2 transcript was about 8–10% of the total ABH1 transcripts in the WT, about 11% in pinin-1, about 15% in acinus-2, but more than 50% in acinus-2 pinin-1 (Fig. 4b, c). Expression of either YFP-AtPININ or AtACINUS-GFP in the acinus-2 pinin-1 background rescued the ABH1 intron retention phenotype (Fig. 4b, c). Consistent with compromised ABH1 activity, the gene expression changes in acinus-2 pinin-1 show a strong correlation to those in abh1, with Spearman’s correlation = 0.74 as calculated by AtCAST3.1 (Supplementary Fig. 8)32,33.

Intron retention in HAB1 has been reported to cause ABA hypersensitive phenotypes34,35. HAB1 did not display any apparent splicing defects in our RNA-seq and RT-PCR analysis of the 12-day-old seedling. However, after ABA treatment, HAB1 intron retention is significantly increased in acinus pinin compared to the WT. While the expression level of HAB1 transcripts was increased similarly in WT and acinus pinin, the WT seedlings maintained relatively similar ratios between different splice forms of HAB1 before and after ABA treatment, whereas the acinus pinin mutant accumulated a much increased level of the intron-containing HAB1.2 and a reduced level of fully spliced HAB1.3 after ABA treatment (Fig. 4d, e). HAB1.2 encodes a dominant negative form of HAB1 protein that activates ABA signaling34,35. Therefore, the accumulation of HAB1.2 should contribute to the ABA hypersensitivity of the acinus pinin mutant.

To test whether AtACINUS is directly involved in AS of ABH1 and HAB1, we carried out an RNA immunoprecipitation (RNA-IP) experiment using an AtACINUS-GFP/acinus-2 transgenic line, with 35S::GFP transgenic plants as the negative control. Immunoprecipitation using an anti-GFP antibody pulled down significantly more ABH1 and HAB1 RNAs in AtACINUS-GFP/acinus-2 than in the 35S::GFP control (Fig. 4f, g), indicating that AtACINUS interacts with ABH1 and HAB1 RNAs in vivo and is involved in their splicing.

AtACINUS regulates flowering through repression of FLC

Consistent with the late-flowering phenotype of acinus pinin (Figs. 1e, 5a), our RNA-seq data showed an increased expression level of the floral repressor FLC, without obvious alteration of the splicing pattern (Supplementary Fig. 9a). The RT-qPCR analysis confirmed the increased levels of FLC RNA that correspond to the severity of the late-flowering phenotypes in the single and double mutants (Fig. 5b). As FLC expression is also controlled by its anti-sense RNA, which undergoes AS36,37, we analyzed the anti-sense FLC RNAs using RT-qPCR. The results showed a dramatic increase of the class I anti-sense RNA and a slight increase of the class II anti-sense RNA of FLC, but no obvious change of the splicing efficiency of the FLC anti-sense RNAs (Supplementary Fig. 9b–d). AtACINUS was recently reported to associate with VAL1 and VAL2, which bind to the FLC promoter to repress transcription26. We thus performed chromatin immunoprecipitation (ChIP) assays to test whether AtACINUS is associated with the FLC locus, and our results show that AtACINUS interacts with the DNA of the promoter and first intron regions but not the 3′ region of FLC in vivo (Fig. 5c). Together our results provide evidence for a role of AtACINUS in regulating the transcription of FLC.

Fig. 5: The acinus-2 pinin-1 double mutant is late flowering with increased FLC expression.
figure 5

a Rosette leaf numbers of WT, acinus-2, pinin-1, and acinus-2 pinin-1 at bolting stage grown in long day condition. Error bars indicate SD calculated from n > 12. Values represent mean ± SD calculated from at least 12 plants (n > 12). Statistically significant differences to WT were determined by two-tailed t test. The P values for a, b, and c are 1.35E−6, 2.48E−3, and 5.15E−9. b FLC expression level relative to UBQ (At5g15400) in WT, acinus-2, pinin-1, and acinus-2 pinin-1, determined by RT-qPCR in 12-day-old seedlings. Values represent mean ± SD calculated from three biological replicates (n = 3). Statistically significant differences to WT were determined by two-tailed t test. The P values for a, b, and c are 5.15E−3, 3.32E−3, and 4.91E−3. c Analysis of AtACINUS-GFP association with the FLC locus by ChIP-PCR in 12-day-old AtACINUS-GFP/acinus-2 seedlings. WT serves as the negative control. Bars below the gene structure diagram represent regions analyzed by PCR (blue bars indicate regions enriched after immunoprecipitation). GFP IP shows PCR products using immunoprecipitated DNA. CO-FACTOR FOR NITRATE, REDUCTASE, AND XANTHINE DEHYDROGENASE 5 (CNX5) serves as an internal control to show non-specific background DNA after immunoprecipitation. PCR reactions were set to 28 cycles.

AtACINUS-dependent AS events are altered in spy and sec

To study how O-linked sugar modification affects the function of AtACINUS, we tested if the AtACINUS-dependent AS events are altered in the spy and sec mutants. Of the ten AtACINUS-dependent intron splicing events we have tested, four showed alterations in the spy mutant and one showed alteration in the sec mutant (Fig. 6).

Fig. 6: A subset of AtACINUS-dependent intron splicing events are affected in the spy and sec mutants.
figure 6

RT-PCR of HAB1, EMB2247, and TRM4D in 7-day-old WT, acinus-2 pinin, spy-t1, spy-4, sec-2, and sec-5 seedlings with primers flanking the targeted introns.

In the 7-day-old light-grown plants, splicing of the 12th intron and the 15th intron of TRNA METHYLTRANSFERASE 4D (TRM4D, At4g26600) was enhanced in the acinus-2 pinin double mutant compared to that in the WT. In the loss-of-function mutants spy-4 and spy-t1 (SALK_090580), the splicing efficiency of these two introns were also enhanced. In contrast, the loss-of-function mutants sec-2 and sec-5 showed an increased retention of the 12th intron (Fig. 6). These results suggest that SPY and SEC have opposite effects on AtACINUS function in TRM4D splicing. The spy-t1 and spy-4 mutants accumulated more HAB1.3 and less HAB1.2 than WT, while acinus-2 pinin accumulated more HAB1.2 than the WT (Fig. 6), consistent with their opposite seed germination phenotypes. In addition, the splicing efficiency of the 14th intron of EMBRYO DEFECTIVE 2247 (Emb2247, AT5G16715) was reduced in the acinus-2 pinin double mutant, but was increased in the spy-t1 and spy-4 mutants compared to WT (Fig. 6). These results support that the O-linked sugar modifications of AtACINUS modulate its functions in AS of specific RNAs.

AtACINUS associates with transcriptional and splicing factors

To understand the molecular mechanisms of AtACINUS function, we conducted two immunoprecipitations followed by mass spectrometry (IP-MS) experiments. In the first experiment, immunoprecipitation was performed in three biological replicates using the AtACINUS-GFP/acinus-2 plants and the anti-GFP nanobody. Transgenic plants expressing a Tandem-Affinity-Purification-GFP (TAP-GFP) protein were used as control38. The proteins co-immunoprecipitated with AtACINUS-GFP were identified based on enrichment (FDR = 0.01, S0 = 2) relative to the TAP-GFP control, quantified by label-free mass spectrometry analysis. In the second experiment, AtACINUS-associated proteins were identified by 15N stable-isotope-labeling in Arabidopsis (SILIA) quantitative mass spectrometry. WT and acinus-2 mutant seedlings were metabolically labeled with 14N and 15N, and immunoprecipitation was performed using the anti-AtACINUS antibody, followed by mass spectrometry analysis. The isotope labels were switched in the two biological replicates. AtACINUS-associated proteins were identified based on enrichment in the WT compared to the acinus mutant control. These IP-MS experiments consistently identified 46 AtACINUS-associated proteins (Fig. 7a, Supplementary Fig. 10a, and Supplementary Data 3). These included SR45 and AtSAP18, supporting the existence of an evolutionarily conserved ASAP complex in Arabidopsis. The AtACINUS interactome also included a large number of proteins homologous to known components of the spliceosome, including five Sm proteins, one protein of the U2 complex, four proteins in the U5 complex, 17 proteins of the nineteen complex (NTC) and NTC-related complex (NTR)39,40,41. In addition, AtACINUS associated with six proteins of the exon junction complex (EJC) core and the EJC-associated TRanscription-EXport (TREX) complex, three proteins of the small nucleolar ribonucleoprotein (snoRNP) complexes, and four other splicing-related proteins (Fig. 7a and Supplementary Data 3)41,42,43,44,45. AtACINUS interactome also included a component of the RNA Polymerase II Associated Factor 1 complex (PAF1C) (Fig. 7a and Supplementary Data 3). The interactome data suggest that, similar to mammalian Acinus, AtACINUS plays dual roles in AS and transcriptional regulation.

Fig. 7: AtACINUS is O-GlcNAc and O-fucose modified and associates with spliceosomal complexes, transcriptional regulators, and chromatin remodeling proteins.
figure 7

a Diagram shows functional groups of AtACINUS-associated proteins. Proteins are grouped in boxes based on their association with known complexes or functions. Positive regulators of FLC are highlighted in red and negative regulators in blue. Seven-day-old seedlings were used for the label-free IP-MS experiments and 14-day-old seedlings were used for the 15N stable-isotope-labeling in Arabidopsis (SILIA) quantitative MS experiments. b, c Higher energy collisional dissociation (HCD) mass spectra show O-GlcNAcylation on Thr79 and a sequence spanning amino acid 400–423 of AtACINUS. The sequence ion series that retain this modification (shifted by 203 Da) are labeled in blue (b). The sequence ion series that have lost the modification are labeled in red. HexNAc oxonium ion (m/z 204) and its fragments masses are labeled in red. d HCD spectrum shows O-fucosylation on a sequence spanning amino acid 169–197 of AtACINUS with neutral loss. e Proposed model of a molecular pathway in which nutrient sensing O-GlcNAcylation and O-fucosylation modulate the evolutionarily conserved RSB-domain protein AtACINUS, which controls transcription and alternative RNA splicing of specific target genes to modulate stress hormone sensitivity and developmental transitions such as seed germination and flowering in plants.

The AtACINUS interactome includes five proteins that are genetically involved in regulating FLC and flowering (Fig. 7a and Supplementary Data 3). These are BRR2 and PRP8 of the U5 complex, ELF8 of the PAF1C, and SR45 and AtSAP18 of the ASAP complex19,37,46,47. These results suggest that AtACINUS may regulate FLC expression through a complex protein network involving multiple regulatory pathways.

We have previously identified O-GlcNAc modification on Thr79 on AtACINUS13 (Fig. 7b) after LWAC enrichment. Mass spectrometry analysis following affinity purification of AtACINUS identified additional O-GlcNAc modification on the peptide containing amino acids 407–423 (Fig. 7c and Supplementary Fig. 10b), as well as O-fucosylation on the peptide containing amino acids 169–197 (Fig. 7d). These results confirm that AtACINUS is a target of both O-GlcNAc and O-fucose modifications.

Using targeted mass spectrometry analysis, we confirmed that the acinus-2 pinin double mutant expressed only the AtACINUS’s N-terminal peptides (at about 20% WT level), but no detectable peptides of the C-terminal region (after T-DNA insertion) (Supplementary Fig 11 and Supplementary Data 5). Both N- and C-terminal peptides of AtPININ were undetectable in the acinus-2 pinin mutant (Supplementary Fig 12 and Supplementary Data 5). Meanwhile, SR45 and AtSAP18 protein levels were dramatically reduced to 3.9% and 2.7% of WT levels, respectively (Supplementary Figs. 13 and 14 and Supplementary Data 5). Together, these results indicate that the stability of the other members of the ASAP and PSAP complexes is dependent on AtACINUS and AtPININ.


Our recent identification of O-GlcNAcylated proteins in Arabidopsis enabled functional study of this important signaling mechanism in plants13. Here our systematic analysis of one of these O-GlcNAcylated proteins, AtACINUS, demonstrates its functions as a target of O-GlcNAc and O-fucose signaling and a component of the evolutionarily conserved ASAP complex that regulates transcription and RNA AS thereby modulating stress responses and developmental transitions. Our comprehensive genetic, transcriptomic, and proteomic analyses provide a large body of strong evidence illustrating a molecular pathway in which nutrient sensing O-GlcNAcylation and O-fucosylation modulate specific functions of the evolutionarily conserved RSB-domain protein AtACINUS to modulate stress hormone sensitivity, seed germination, and flowering in plants (Fig. 7e).

Studies in animals have identified Acinus and Pinin as essential cellular components that bridge chromatin remodeling, transcription, and splicing through the formation of analogous ASAP and PSAP complexes14,16,17,48,49,50,51. Sequence alignment and phylogenetic analysis show that the Arabidopsis orthologs, AtACINUS and AtPININ, share higher levels of sequence similarity to their animal counterparts than to each other and appear to have evolved independently since the separation of the plant and metazoan kingdoms14. Considering their evolutionary distance and limited sequence similarity (12 amino acid residues in the RSB motif), it was surprising that the functions of AtACINUS and AtPININ are genetically redundant. This represents likely the least sequence similarity between two redundant genes and raises cautions for prediction of genetic redundancy based on the level of sequence similarity.

The developmental functions in seed germination and flowering seem to involve AtACINUS’s distinct activities in splicing and transcription of key components of the regulatory pathways. Specifically, AS events in ABH1 and HAB1 are likely the major mechanisms by which AtACINUS modulates ABA signaling dynamics to control seed germination and stress responses. ABH1 is an mRNA cap-binding protein that modulates early ABA signaling30,31. The loss-of-function abh1 mutant with a T-DNA insertion in the 8th intron is ABA hypersensitive with enhanced early ABA signaling30. Similarly, the retention of the 10th intron of ABH1 in acinus pinin mutant is expected to truncate its C-terminal half and cause loss of ABH1 function and thus increase of ABA sensitivity. Supporting the functional role of the ASAP/PSAP-ABH1 pathway, we observed a significant correlation between the transcriptomic changes in abh1 and the acinus pinin double mutant (Supplementary Fig. 8)32,33. A recent proteomic study showed that the ABH1 protein level was decreased in the sr45 mutant23, whereas a reduction of ABH1 RNA level to ~30% caused obvious phenotypes in potato52.

AtACINUS-mediated AS of HAB1 switches a positive feedback loop to a negative feedback loop in the ABA signaling pathway. HAB1 encodes a phosphatase that dephosphorylates the SNF1-related protein kinases (SnRK2s) to inhibit ABA responses, and the ligand-bound ABA receptor inhibits HAB1 to activate ABA responses53,54. The intron-containing HAB1.2 encodes a dominant negative form of HAB1 protein that lacks the phosphatase activity but still competitively interacts with SnRK2, thus activating, instead of inhibiting, ABA signaling34,35. As ABA signaling feedback increases the HAB1 transcript level, the AtACINUS-mediated AS switches a positive feedback loop that reinforces ABA signaling to a negative feedback loop that dampens ABA signaling. Such a switch is presumably important for the different ABA signaling dynamics required for the onset of and recovery from stress responses or dormancy.

The relative contributions of intron retention of ABH1 and HAB1 to ABA sensitivity will need to be quantified by genetic manipulation of each splicing event. Additional mechanisms may contribute to the ABA-hypersensitivity phenotypes of acinus pinin. For example, the level of SR45 is significantly decreased in acinus pinin, while loss of SR45 has been reported to cause accumulation of SnRK1 which is a positive regulator of stress and ABA responses55.

The late-flowering phenotype of the acinus pinin mutant correlated with increased FLC expression. A role of AtACINUS in repressing FLC has been suggested based on its association with the VAL1 transcription factor, which binds to the FLC promoter26. Our results provide genetic evidence for the function of AtACINUS in repressing FLC expression. Further, our ChIP-PCR analysis shows that AtACINUS associates with genomic DNA of the promoter region and the first intron of FLC, confirming a direct role in transcriptional regulation of FLC. These results provide critical evidence for the hypothesis that the AtACINUS represses FLC by AtSAP18-mediated recruitment of the Sin3 histone deacetylase complex (HDAC)26. It is worth noting that overexpression of AtSAP18 in the sr45 mutant increased FLC expression and further delayed flowering23. It is possible that the transcriptional repression function of AtSAP18 requires the ASAP/PSAP complex. It is also worth noting that the AtACINUS interactome includes several proteins known to be involved in regulating FLC expression and flowering. Among these, BRR2 and PRP8 are components of the U5 complex and mediate splicing of the sense and anti-sense transcripts of FLC to inhibit and promote flowering, respectively37,46. ELF8 is a component of the PAF1 complex and promotes histone methylation of FLC chromatin47. The identification of additional FLC-regulators as AtACINUS-associated proteins suggests that AtACINUS may regulate FLC expression through complex protein networks. Genetic evidence supports that ELF8/PAF1C and SR45 also have dual functions in regulating FLC expression and ABA responses18,19,22,56, suggesting that the functions of AtACINUS in seed germination and flowering may involve overlapping protein networks.

Structural studies in metazoan systems showed that the RSB domains of Acinus and Pinin directly interact with RNPS1 and SAP18, forming a ternary ASAP and PSAP complexes that have both RNA- and protein-binding properties as well as abilities to interact with both RNA splicing machinery and histone modifiers14. ASAP and PSAP function as EJC peripheral protein complexes to modulate RNA processing15,57. Our quantitative proteomic analyses of the AtACINUS interactome provide strong evidence for interaction with SR45 (ortholog of RNPS1) and AtSAP18, as well as components of EJC and additional splicing factors. However, some proteins, such as SPY and SEC, may interact transiently and were not detected by IP-MS. While our proteomic data do not distinguish the proteins that directly interact with AtACINUS from those that associate indirectly as a subunit of the interacting protein complexes, the greatly reduced levels of SR45 and AtSAP18 proteins in acinus pinin are consistent with the direct interactions predicted based on the conserved RSB domain. Similarly, the sr45 mutation leads to a near absence of AtSAP18 and a mild decrease of the AtACINUS protein level in the inflorescence tissues23. Together these observations support the notion that AtACINUS and AtPININ mediate formation of similar ASAP and PSAP complexes and stabilize SR45 and AtSAP18 in plants.

Studies in human cells have shown that Acinus and Pinin mediate splicing of distinct RNAs and that Acinus cannot rescue the splicing defects caused by knockdown of Pinin15. In contrast, AtACINUS and AtPININ appear to have largely redundant and interchangeable functions. It is possible that both AtACINUS and AtPININ, through their RSB domain, recruit SR45 and AtSAP18, which determine target specificities. However, AtACINUS and AtPININ may have subtle differences in their functions. Like human Acinus, AtACINUS contains two additional conserved domains that are absent in AtPININ. Further, the regions of AtACINUS and AtPININ, as well as human Acinus and Pinin, outside the RSB domain contain mostly divergent intrinsically disordered sequences58 (Supplementary Fig. 15). These distinct sequences may provide specificity in interactions with target transcripts and partner proteins or in regulation by PTMs58. Indeed, O-GlcNAcylated residues (Thr79 and amino acids 407–423) and the O-fucosylated site (amino acids 169–197) were in the intrinsically disordered regions of AtACINUS, whereas no O-GlcNAc or O-fucose modification was detected in AtPININ, though this could be due to partial sequence coverage of our mass spectrometry analysis. Deep RNA-seq analysis with higher sequence coverage of the single and double mutants of acinus and pinin will be required to fully understand their functional overlap and specificities.

How SEC/O-GlcNAc and SPY/O-fucose modulate development and physiology of plants is not fully understood at the molecular level. The mechanism of regulating GA signaling involves antagonistic effects of O-fucosylation and O-GlcNAcylation of the DELLA proteins10. Similarly, we observed the opposite effects of spy and sec on the splicing of the 12th intron of TRM4D, suggesting distinct effects of O-GlcNAcylation and O-fucosylation on AtACINUS functions. Consistent with their different phenotype severities, more AS events were affected in spy than sec. The spy mutant showed increased splicing for four of the ten introns analyzed; two of these introns (in TRM4D) were more spliced and the other two (HAB1 and EMB2247) were less spliced in the acinus pinin mutant than in WT, suggesting that the SPY-mediated O-fucosylation may have different effects on AtACINUS activities on different transcripts. The two O-GlcNAc-modified residues (Thr79 and amino acids 407–423) and the O-fucose modified residue (amino acids 169–197) are in different regions of the intrinsically disordered sequence58 (Supplementary Fig. 15), suggesting that PTMs in the disordered sequences play roles in substrate-specific splicing activities.

The high percentage of AtACINUS-dependent AS events affected in spy and sec supports an important function of AtACINUS in mediating the regulation of AS by O-glycosylation. On the other hand, AtACINUS-independent mechanisms may also contribute to the regulation, as the O-GlcNAcylated Arabidopsis proteins include additional RNA-binding and splicing factors13, such as SUS2 which is in the AtACINUS interactome. Deep transcriptomic analysis of spy, sec, and conditional double spy sec mutants will be required to better understand how O-GlcNAc and O-fucose modulate RNA processing and AtACINUS function. Genetic analyses have suggested that SPY acts upstream of the ABA insensitive 5 (ABI5) transcription factor in regulating seed germination8. The molecular link between SPY/O-fucose and ABA signaling has remained unknown. Our results support a hypothesis that O-fucose modification modulates AtACINUS activity in splicing a subset of transcripts including HAB1 to modulate ABA sensitivity. The biological function of this SPY-AtACINUS pathway remains to be further evaluated by genetic analyses including mutagenesis of the O-fucosylation sites of AtACINUS. It is likely that parallel pathways also contribute to the regulation of ABA sensitivity and seed germination by O-fucosylation and O-GlcNAcylation. For example, increased GA signaling was thought to contribute to ABA hyposensitivity in the spy mutant59. Further, the ABA response element binding factor 3 (ABF3) is also modified by O-GlcNAc13. The function of O-glycosylation in stress responses seems to be conserved, as large numbers of molecular connections between O-GlcNAc and stress response pathways have been reported in metazoans5.

How O-linked glycosylation of AtACINUS affects its transcriptional activity at the FLC locus remains to be investigated. Both spy and sec mutants flower early, opposite to acinus pinin. While spy shows a strong early flowering phenotype, the FLC expression level was unaffected in spy under our experimental conditions (Supplementary Fig. 16), suggesting that SPY regulates flowering independent of FLC. The FLC level was decreased in sec60, supporting the possibility that O-GlcNAcylation affects AtACINUS transcription activity. However, the effect of sec on FLC expression could also be mediated by other O-GlcNAc-modified flowering regulators13,60.

Our study reveals important functions of AtACINUS in developmental transitions and a previously unknown function of O-linked glycosylation in regulating RNA AS. While we were getting our revised manuscript ready for submission, evidence was reported for a similar function of O-GlcNAc in intron splicing in metazoan and for broad presence of stress-dependent intron retention in plants. Interestingly, inhibition of OGT was found to increase splicing of detained introns in human cells61. Detained introns are a novel class of post-transcriptionally spliced (pts) introns, which are one or few introns retained in transcripts where other introns are fully spliced62. Transcripts containing pts introns are retained on chromatin and are considered a reservoir of nuclear RNA poised to be spliced and released when a rapid increase of protein level is needed, such as in neuronal activities62,63. A recent study uncovered a large number of pts introns in Arabidopsis. A significant portion of these pts introns show enhanced intron retention under stress conditions. Several splicing factors involved in pts intron splicing, MAC3A, MAC3B, and SKIP64, are parts of the AtACINUS interactome. Among the introns retained in the acinus pinin mutant, 114 are pts introns, which is about 1.7-fold the random probability (p value <3.0E−9). These pts introns include the intron retained in ABH1 but not that in HAB1, consistent with translation of the dominant negative form of HAB1.2 (refs. 34,35). Together with these recent developments, our study raises the possibility that AtACINUS plays important roles in the splicing of pts introns, acting downstream of the metabolic signals transduced by SPY/O-fucose and SEC/O-GlcNAc. Our study supports an evolutionarily conserved function of O-glycosylation in regulating RNA splicing, thereby linking metabolic signaling with switches of cellular status between normal and stress conditions as well as during developmental transitions.


Plant materials

All the Arabidopsis thaliana plants used in this study were in the Col-0 ecotype background. The plants were grown in greenhouses with a 16-h light/8-h dark cycle at 22–24 °C for general growth and seed harvesting. For seedlings grown on the medium in Petri dishes, the sterilized seeds were grown on ½ Murashige and Skoog (MS) medium and supplemented with 0.7% (w/v) phytoagar. Plates were placed in a growth chamber under the constant light condition at 21–22 °C. T-DNA insertional mutants atacinus-1 (Salk_078554, insertion position +1744 relative to the genomic translational start site of At4G39680), atacinus-2 (WiscDsLoxHs108_01G, insertion position +674), atpinin-1 (GABI_029C11/CS402723, insertion position +1817 of At1G15200), spy-t1 (Salk_090580), and sec-5 (Salk_034290) were obtained from Arabidopsis Biological Resource Center. The spy-4 and sec-2 seeds that have been backcrossed to Columbia for six generations were provided by Neil Olszewski lab.

Germination assay

Seeds were surface sterilized with 70% (v/v) ethanol and 0.1% (v/v) Triton X-100 sterilization solution for 5 min. The sterilization solution was then removed and seeds were resuspended in 100% ethanol and dried on a filter paper. The sterilized seeds were then plated on ½ MS medium supplemented with mock or ABA. The seeds were placed in 4 °C cold room for 3 days for stratification before moving into a growth chamber to germinate. Germination was defined as obvious radicle emergence from the seed coat.

Gene cloning and plant transformation

The AtACINUS cDNA was initially cloned into the vector pENTR-D/TOPO and subsequently into the binary vector pGWB5 to generate the 35S::AtACINUS-GFP plasmid. The 35S::AtACINUS-GFP binary plasmid was transformed into acinus-2 plants by floral dipping with Agrobacterium tumefaciens strain GV3101. A homozygous 35S::AtACINUS-GFP/acinus-2 plant was selected for similar protein expression level to the endogenous AtACINUS protein of WT plants using a native α-AtACINUS antibody, and crossed with acinus-2 pinin-1 to obtain 35S::AtACINUS-GFP/acinus-2 pinin-1 transgenic lines. Similarly, 35S::AtACINUS-YFP-TurboID plasmid was generated by LR reaction of gateway-compatible 35S::YFP-TbID65 with pENTR-AtACINUS and transformed to acinus-2 pinin-1 to obtain transgenic lines.

The AtPININ cDNA was acquired from the Arabidopsis stock center and subsequently cloned into the binary vector pEarleyGate104 to generate the 35S::YFP-AtPININ vector. The 35S::YFP-AtPININ binary plasmid was transformed into acinus-2 pinin-1/+ plants by floral dipping with A. tumefaciens strain GV3101. Transgenic plants were genotyped for pinin-1 allele to obtain 35S::YFP-AtPININ/acinus-2 pinin-1 transgenic lines.

Bioinformatics analysis

Dendrogram of AtACINUS and AtPININ homologs in different species was constructed using the “simple phylogeny” web tool of EMBL-EBI website with UPMGA method using default settings ( The protein alignment was generated using MUSCLE from EMBL-EBI website with the default setting (,67. Pairwise protein sequence alignment was performed with Blastp from the NCBI blastp suite with E-value set to 0.01. (

Protein disorderliness was predicted based on amino acid sequences using PrDOS ( with the default setting68.

Gene expression correlation was analyzed with AtCAST3.1 using default settings (

RNA sequencing and data analysis

RNA was extracted from 14-day-old WT and acinus-2 pinin-1 seedlings using RNeasy mini kit (Qiagen) and treated with TURBO DNA-free Kit (Ambion) to remove any genomic DNA contamination. The mRNA libraries were constructed using NEBNext RNA Library Prep Kit for Illumina following the standard Illumina protocol. Illumina sequencing was performed in the Sequencing Center for Personalized Medicine, Department of Genetics at Stanford University, using an Illumina HiSeq 2000 System. The RNA-seq data have been deposited at the NCBI Gene Expression Omnibus (GEO) database under the accession number GSE110923.

Differential gene expression was analyzed using STAR and Deseq2. Trimmed and quality control-filtered sequence reads were mapped to the Arabidopsis reference genome (TAIR10) using STAR (v.2.54) in two pass mode (parameters: –outFilterScoreMinOverLread 0.3, –outFilterMatchNminOverLread 0.3, –outSAMstrandField intronMotif, –outFilterType BySJout, –outFilterIntronMotifs RemoveNoncanonical, –quantMode TranscriptomeSAM GeneCounts)69. To obtain uniquely mapping reads, these were filtered by mapping quality (q20), and PCR duplicates were removed using Samtools rmdup (v.1.3.1). Gene expression was analyzed in R (v.3.4.1) using DEseq2 (v.1.16.1)70. Significant differentially expressed genes are selected based on adj p-value <0.02 and fold change >2.

AS analysis was performed with RACKJ using default setting (online manual available at Raw intron retention data were analyzed and filtered to reduce false positives with two criteria: (1) fold change of intron retention >2, p-value <0.05 in a two-tailed t-test and (2) intron RPKM >1 and estimated percentage of IR >5% in the sample that shows increased IR in the intron. Raw exon skipping (ES) data were analyzed and filtered with two criteria: (1) fold change of ES rate >2, p-value <0.05 in a two-tailed t-test, and (2) increased ES event is supported by reads with RPKM >1 and ES rate >5%. For alternative donor/acceptor usage discovery, only events that appear significantly different in each pair-wise comparison between WT and acinus-2 pinin-1 (Fisher’s exact test p-value <0.05) were considered significant and were further filtered with two criteria: (1) fold change >2 and (2) increased alternative donor/acceptor usage is supported by reads with RPKM >1 and rate >5%.

RNA extraction, reverse transcription PCR

RNA was extracted from seedlings using Spectrum™ Plant Total RNA Kit (Sigma) and treated with TURBO DNA-free Kit (Ambion) to remove any genomic DNA contaminants. Purified RNA (500 ng) is subjected to cDNA synthesis using RevertAid Reverse Transcriptase (Thermo) with Oligo(dT)18 primer. The synthesized cDNA was used for PCR and qPCR analyses. PCR products were analyzed by gel electrophoresis and the PCR band intensities were quantified using ImageJ. The qPCR analyses were performed with the SensiMix™ SYBR® & Fluorescein Kit (Bioline) on a LightCycler 480 (Roches). For each sample, two technical replicates were performed. The comparative cycle threshold method was used for calculating transcript level. Primers used for FLC antisense analysis are the same as in the previous publication37. Sequences of oligo primers are listed in Supplementary Data 4.

RNA immunoprecipitation

RNA-IP was performed using a protocol modified based on published procedures22. Briefly, 3 g of tissues of 7-day-old 35S::AtACINUS-GFP/acinus-2 and 35S::GFP seedlings were cross-linked with 1% (v/v) formaldehyde for 15 min. Cross-linked RNA–protein complexes were extracted in NLB buffer (20 mmol/L Tris-HCl, pH 8.0, 150 mmol/L NaCl, 2 mmol/L EDTA, 1% (v/v) Triton X-100, 0.1% (w/v) SDS, 1 mmol/L PMSF, and 2× Protease Inhibitor (Roche)) and sheared by sonication (25% amplitude, 0.5 s on/0.5 s off for 2 min × 3 cycles on a Branson Digital Sonifier). Immunoprecipitation was carried out with Protein A magnetic beads (Thermo Fisher) that were pre-incubated overnight with homemade anti-GFP antibody (5 µg for each sample) for 1 h on a rotator. Beads were washed five times with 1 mL of NLB buffer (no SDS, 0.5% (v/v) Triton X-100) with 80 U/mL RNase inhibitor. To elute the immuno-complex, 100 µL of elution buffer (20 mmol/L Tris-HCl, pH 8.0, 10 mmol/L EDTA, 1% (w/v) SDS, 800 U/mL RNase inhibitor) was added to the beads and incubated at 65 °C for 15 min. The elute was incubated with 1 µL of 20 mg/mL Protease K at 65 °C for 1 h for protein digestion and reverse-crosslinking. RNA was purified and concentrated using the RNA Clean & Concentrator™ kit (Zymo). On-column DNase digestion was performed to remove DNA contaminations. Samples were kept on ice whenever possible during the experiment. Three biological replicates were performed and the co-immunoprecipitated RNAs were quantified with RT-qPCR, and the results were normalized to PP2a and 25S rRNA72.


ChIP analysis was performed using a similar protocol to the previous publications73. Briefly, tissue crosslinking, protein extraction, and immunoprecipitation were carried out as described above for RNA-IP. The beads were washed with low-salt buffer (50 mmol/L Tris-HCl at pH 8.0, 2 mmol/L EDTA, 150 mmol/L NaCl, and 0.5% (v/v) Triton X-100), high-salt buffer (50 mmol/L Tris-HCl at pH 8.0, 2 mmol/L EDTA, 500 mmol/L NaCl, and 0.5% (v/v) Triton X-100), LiCl buffer (10 mmol/L Tris-HCl at pH 8.0, 1 mmol/L EDTA, 0.25 mol/L LiCl, 0.5% (w/v) NP-40, and 0.5% (w/v) sodium deoxycholate), and TE buffer (10 mmol/L Tris-HCl at pH 8.0 and 1 mmol/L EDTA), and eluted with elution buffer (1% (w/v) SDS and 0.1 mmol/L NaHCO3). After reverse cross-linking and proteinase K digestion, the DNA was purified with a PCR purification kit (Thermo Fisher) and analyzed by PCR. Three biological replicates were performed. FLC primers were based on the previous publications47.

SILIA-MS quantitative analysis of the AtACINUS interactome

Stable-isotope-labeling in Arabidopsis mass spectrometry (SILIA-MS) was used for quantitative analysis of the AtACINUS interactome. The WT and acinus-2 plants were grown for 2 weeks at 21 °C under constant light on vertical plates of 14N or 15N medium (Hogland’s No. 2 salt mixture without nitrogen 1.34 g/L, 6 g/L Phytoblend, 2 µmol/L propiconazole, and 1 g/L KNO3 or 1 g/L K15NO3 (Cambridge Isotope Laboratories), pH 5.8). About 5 g of tissue was harvested for each sample, ground in liquid nitrogen and stored at −80 °C. Immunoprecipitation was performed as described previously with slight modifications74. Briefly, proteins were extracted in 10 mL of MOPS buffer (100 mmol/L MOPS, pH 7.6, 150 mmol/L NaCl, 1% (v/v) Triton X-100, 1 mmol/L phenylmethylsulfonyl fluoride (PMSF), 2× Complete protease inhibitor cocktail, and PhosStop cocktail (Roche)), centrifuged, and filtered through two layers of Miracloth. The flow through was incubated with 20 µg of anti-AtACINUS antibody for 1 h at 4 °C, then 50 µL of protein A agarose beads were added and incubated for another hour, followed by four 2-min washes with immunoprecipitation buffer. At the last wash, 14N-labeled WT and 15N-labeled acinus-2 IP samples or reciprocal 15N-labeled WT and 14N-labeled acinus-2 IP samples were mixed, and eluted with 2× SDS buffer. The eluted proteins were separated by SDS-PAGE. After Coomassie Brillant blue staining, the whole lane of protein samples was excised in ten segments and subjected to in-gel digestion with trypsin.

The peptide mixtures were desalted using C18 ZipTips (Millipore) and analyzed on an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher), equipped with a NanoAcquity liquid chromatography system (Waters). Peptides were loaded onto a trapping column (NanoAcquity UPLC 180 µm × 20 mm; Waters) and then washed with 0.1% (v/v) formic acid. The analytical column was a BEH130 C18 100 µm × 100 mm (Waters). The flow rate was 600 nL/min. Peptides were eluted by a gradient from 2–30% solvent B (100% (v/v) acetonitrile/0.1% (v/v) formic acid) over 34 min, followed by a short wash at 50% solvent B. After a precursor scan was measured in the Orbitrap by scanning from mass-to-charge ratio 350 to 1500, the six most intense multiply charged precursors were selected for collision-induced dissociation in the linear ion trap.

Tandem mass spectrometry peak lists were extracted using an in-house script PAVA, and data were searched using Protein Prospector against the Arabidopsis Information Resource (TAIR10) database, to which reverse sequence versions were concatenated (a total of 35,386 entries) to allow estimation of a false discovery rate (FDR). Carbamidomethylcysteine was searched as a fixed modification and oxidation of methionine and N-terminal acetylation as variable modifications. Data were searched with a 10 p.p.m. tolerance for precursor ions and 0.6 Da for fragment ions. Peptide and protein FDRs were set as 0.01 and 0.05. 15N-labeled amino acids were also searched as a fixed modification for 15N data. 15N labeling efficiency was calculated as about 96%, by manually comparing experimental peak envelope data of the 15N-labeled peptide from top 10 proteins in the raw data to theoretical isotope distributions using Software Protein-prospector (MS-Isotope app). Quantification was done using Protein Prospector which automatically adjusts the L/H ratio with labeling efficiency. The SILIA ratio (WT/acinus-2) was normalized using the average ratios of non-specific interactor ribosomal proteins (with more than five peptides). 15N labeling samples in general have lower identification rates of proteins because of incomplete (96%) labeling efficiency. The data have been deposited to PRIDE with project accession: PXD020700.

Label-free mass spectrometric analysis of AtACINUS and its interactome

The AtACINUS-GFP/acinus-2 and TAP-GFP seedlings38 were grown for 7 days at 21 °C under constant light on ½ MS medium. Tissues were harvested, ground in liquid nitrogen, and stored at −80 °C.

Immunoprecipitation was performed as described previously with slight modifications74. Briefly, proteins were extracted in MOPS buffer (100 mmol/L MOPS, pH 7.6, 150 mmol/L NaCl, 1% (v/v) Triton X-100, 1 mmol/L PMSF, 2× Complete protease inhibitor cocktail, and PhosStop cocktail (Roche)) and 20 µmol/L PUGNAc inhibitor (Sigma), centrifuged, and filtered through two layers of Miracloth, then incubated with a modified version of LaG16-LaG2 anti-GFP nanobody75 conjugated to Dynabeads (Invitrogen), for 3 h at 4 °C, followed by four 2-min washes with immunoprecipitation buffer and eluted with 2% (w/v) SDS buffer containing 10 mmol/L tris(2-carboxyethyl) phosphine (TCEP) and 40 mmol/L chloroacetamide at 95 °C for 5 min. The eluted proteins were separated by SDS-PAGE. After Colloidal blue staining, the whole lane of protein samples was excised in two segments and subjected to in-gel digestion with trypsin. Three biological experiments were performed.

The peptide mixtures were desalted using C18 ZipTips (Millipore) and analyzed on a Q-Exactive HF hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher) equipped with an Easy LC 1200 UPLC liquid chromatography system (Thermo Fisher). Peptides were separated using analytical column ES803 (Thermo Fisher). The flow rate was 300 nL/min and a 120-min gradient was used. Peptides were eluted by a gradient from 3 to 28% solvent B (80% (v/v) acetonitrile/0.1% (v/v) formic acid) over 100 min and from 28 to 44% solvent B over 20 min, followed by a short wash at 90% solvent B. Precursor scan was from mass-to-charge ratio (m/z) 375 to 1600 and top 20 most intense multiply charged precursors were selected for fragmentation. Peptides were fragmented with higher-energy collision dissociation (HCD) with normalized collision energy (NCE) 27.

The raw data were processed by MaxQuant using most of the preconfigured settings76. The search was against the same TAIR database as mentioned above. Carbamidomethylcysteine was searched as a fixed modification and oxidation of methionine and N-terminal acetylation as variable modifications. Data were searched with a 4.5 p.p.m. tolerance for precursor ion and 20 p.p.m. for fragment ions. The second peptide feature was enabled. A maximum of two missed cleavages was allowed. Peptide and protein FDRs were set as 0.01. Minimum required peptide length was seven amino acids. Multiplicity was set to 1. Label-free quantification (LFQ) was enabled. The match between runs option was enabled with a match time window of 0.7 min and an alignment time window of 20 min. Quantification was done on unique and razor peptides and a minimum ratio count was set to 2.

The proteinGroups.txt file generated by MaxQuant were loaded to Perseus77. The results were filtered by removing identified proteins by only modified sites, or hits to reverse database and contaminants. LFQ intensity values were logarithmized. The pull-downs were divided to AtACINUS-GFP and TAP-GFP control. Samples were grouped in triplicates and identifications were filtered for proteins having at least three values in at least one replicate group. Signals that were originally zero were imputed with random numbers from a normal distribution (width 0.3, shift = 1.8). Volcano plot was performed with x axis representing the logarithmic ratios of protein intensities between AtACINUS-GFP and TAP-GFP. The hyperbolic curve that separates AtACINUS specific interactor from background was drawn using threshold value FDR 0.01 and curve bend S0 value 2.

LFQ data and SILIA data were combined and filtered to get a high-confidence list of interactors: (1) Significant enrichment in LFQ three biological replicates (FDR = 0.01, S0 = 2); (2) Enrichment of over twofolds in both SILIA biological experiment; or over twofolds in one SILIA experiment, but not identified in second SILIA experiment. If the proteins are only identified and quantified by LFQ three biological replicates, then a higher stringency cut off (enrichment >16-fold, t-value >4) is used. The data were deposited to PRIDE with project accession: PXD020748.

For affinity purification of AtACINUS using in vivo biotinylation, the acinus pinin mutant was transformed with a T-DNA construct that expresses AtACINUS as a fusion with TurboID from the 35S promoter65. The AtACINUS-YFP-TurboID/acinus-2 pinin-1 seedlings were treated with 0 or 50 µmol/L biotin for 3 h. The AtACINUS-YFP-Turbo protein was affinity purified using streptavidin beads as previously described65 using a modified extraction buffer containing 20 µmol/L PUGNAC and 1× PhosphoStop. After on-bead tryptic digestion, the samples were analyzed as described above in the label-free IP-MS section on a Q-Exactive HF instrument. Data were searched as described above but allowing additional modifications: O-GlcNAcylation modification on S/T and neutral loss, O-fucosylation on S/T and neutral loss, phosphorylation on S/T and biotinylation on lysine. The data were deposited to PRIDE with accession number: PXD020749.

Targeted quantification comparing WT and the acinus-2 pinin-1 double mutant

The WT and acinus-2 pinin-1 plants were grown on Hoagland medium containing 14N or 15N (1.34 g/L Hogland’s No. 2 salt mixture without nitrogen, 6 g/L Phytoblend, and 1 g/L KNO3 or 1 g/L K15NO3 (Cambridge Isotope Laboratories), pH 5.8). Proteins were extracted from six samples (one 14N-labeled Col, two of 15N-labeled Col, two of 14N-labeled acinus-2 pinin-1, and one 15N-labeled acinus-2 pinin-1) individually using SDS sample buffer and mixed as the following: one forward sample F1 (14N Col/15N acinus-2 pinin-1) and two reverse samples R2 and R3 (14N acinus-2 pinin-1/15N Col) and separated by the SDS-PAGE gel with a very short run. Two segments (upper part (U) ranging from the loading well to ~50 KD; lower part (L) ranging from ~50 KD to the dye front) were excised, trypsin digested, and analyzed by liquid chromatography mass spectrometry (LC-MS) as described above in the label-free IP-MS section on a Q-Exactive HF instrument using an ES803A analytical column. Data-dependent acquisition was used first to get the peptide information from multiple proteins with peptide mass/charge (m/z), retention time, and MS2 fragments. PININ peptide information was from an IP-MS experiment. For targeted analysis, parallel reaction monitoring (PRM) acquisition78 using a 20-min window was scheduled with an orbitrap resolution at 60,000, AGC value 2e5, and maximum fill time of 200 ms. The isolation window for each precursor was set at 1.4 m/z unit. Data processing was similar to the previous report79 with a 5-p.p.m. window using skyline from 14N- and 15N-labeled samples. Peak areas of fragments were calculated from each sample, the sum of peak areas from the upper gel segment and the lower gel segment was used to calculate the acinus-2 pinin-1/Col ratios for each peptide, normalized to TUBULIN2 to get the normalized ratios. The median number of multiple ratio measurements is used for each protein.

Statistics and reproducibility

Figure 4b,  d and Supplementary Fig. 16 show representative results from two independent experiments, each with three biological repeats. Figure 6 and Supplementary Fig. 3a, b show representative results from two independent experiments. Figure 5c shows results from one experiment with three biological repeats.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.