CAX-INTERACTING PROTEIN4 depletion causes early lethality and pre-mRNA missplicing in Arabidopsis Authors: Uri Aceituno-Valenzuela, Sara Fontcuberta-Cervera, Rosa Micol-Ponce, Raquel Sarmiento-Mañús, Alejandro Ruiz-Bayón, María Rosa Ponce Published: 2024 ABSTRACT AbstractZinc knuckle (ZCCHC) motif-containing proteins are present in unicellular and multicellular eukaryotes, and most ZCCHC proteins with known functions participate in the metabolism of various classes of RNA, such as mRNAs, ribosomal RNAs, and microRNAs. The Arabidopsis (Arabidopsis thaliana) genome encodes 69 ZCCHC-containing proteins; however, the functions of most remain unclear. One of these proteins, CAX-INTERACTING PROTEIN 4 (CXIP4, encoded by AT2G28910), has been classified as a PTHR31437 family member. This family includes human Splicing regulatory glutamine/lysine-rich protein 1 (SREK1)-interacting protein 1 (SREK1IP1), which is thought to function in pre-mRNA splicing and RNA methylation. Metazoan SREK1IP1-like and plant CXIP4-like proteins only share a ZCCHC motif, and their functions remain almost entirely unknown. Here, we studied two loss-of-function alleles of Arabidopsis CXIP4: cxip4-1 is likely null and shows early lethality, and cxip4-2 is hypomorphic and viable, with pleiotropic morphological defects. The cxip4-2 mutant exhibited deregulation of defense genes and upregulation of transcription factor genes, some of which might explain its developmental defects. The cxip4-2 mutant also exhibited increased intron retention events, being more evident in cxip4-1. The specific functions of misspliced genes, such as those involved in “gene silencing by DNA methylation” and “mRNA polyadenylation factor” suggest that CXIP4 has additional functions. In cxip4-2 plants, polyadenylated RNAs accumulate in the nucleus; these could be misspliced mRNAs. The CXIP4 protein localizes to the nucleus in a pattern resembling nuclear speckles rich in splicing factors. Therefore, CXIP4 is required for plant development and survival and mRNA maturation. Introduction Zinc knuckles are 18-residue CCHC-type zinc finger (ZCCHC) motifs with the conserved sequence CX2CX4HX4C. These motifs are present in proteins of unicellular and multicellular eukaryotes, from yeast to plants and animals, but are not found in bacteria. Most ZCCHC-containing proteins with known functions are involved in the metabolism of different classes of RNAs, including messenger RNAs (mRNAs), ribosomal RNAs (rRNAs), and microRNAs (miRNAs). The Arabidopsis (Arabidopsis thaliana) genome encodes 69 ZCCHC-containing proteins, only a few of which have been functionally characterized at some level (reviewed in Aceituno-Valenzuela et al. 2020). The AT2G28910 Arabidopsis gene encodes the ZCCHC-containing protein CAX-INTERACTING PROTEIN 4 (CXIP4). Arabidopsis CXIP4 activated the H+/Ca2+ antiporter CAX1 when expressed in yeast (Cheng et al. 2004). CXIP4 has been classified as a member of the Splicing regulatory glutamine/lysine-rich protein 1 (SREK1)-interacting protein 1 (SREK1IP1) family, which has rarely been studied. Human SREK1IP1 (also known as P18SRP or SFRS12IP1) interacts with the serine-arginine (SR)-rich splicing regulatory protein SREK1, as revealed in yeast two-hybrid (Y2H) assays. In turn, SREK1 interacts with other SR proteins, modulating splice site (SS) selection during alternative splicing (AS) events (Heese et al. 2004). Downregulation of human SREK1IP1 by RNA interference promoted cell proliferation, migration, and invasion in tumor cell lines (Akiyama et al. 2016), pointing to its role in suppressing tumor growth. Human SREK1IP1 has also been identified as an interactor of METHYLTRANSFERASE LIKE 16 (METTL16; Covelo-Molares et al. 2021), one of the two enzymes that catalyze the deposition of N6-methyladenosine (m6A) epitranscriptomic marks onto mRNAs and the U6 small nuclear RNA (snRNA) (Warda et al. 2017). Dominant frameshift alleles of SREK1IP1 have been associated with congenital anosmia (Kamarck et al. 2024). We previously identified Arabidopsis CXIP4 in a Y2H-based screen for interactors of MORPHOLOGY OF ARGONAUTE1-52 SUPPRESSED2 (MAS2; Sánchez-García et al. 2015), which is the putative Arabidopsis ortholog of human NF-kappa-B-activating protein (NKAP; Chen et al. 2003). The multifunctional protein NKAP was recently identified as an exon ligation factor in the post-catalytic (C*) spliceosome (Fica et al. 2019), and as an m6A epitranscriptomic mark reader during miRNA and mRNA maturation (Zhang et al. 2019; Sun et al. 2022). Overexpression of Arabidopsis CXIP4 conferred resistance to dithiothreitol (DTT) and tunicamycin (TM), two drugs that induce endoplasmic reticulum stress and trigger the unfolded protein response (UPR; Hossain et al. 2016). The wheat (Triticum aestivum) putative CXIP4 ortholog (TaCAXIP4) has been proposed to participate in calcium-mediated plant immune responses during Fusarium infection (Chen et al. 2022). More recently, TaCAXIP4 has been found in a complex with several factors involved in the control of AS, including SR-rich proteins (He et al. 2024). Despite these findings, the precise functions of plant CXIP4-like and metazoan SREK1IP1-like proteins remain elusive, and loss-of-function mutants have not been characterized in any eukaryote. Here, we performed functional analysis of Arabidopsis CXIP4 based on two T-DNA insertional alleles: cxip4-1 and cxip4-2. Our findings provide molecular and genetic evidence for the essential role of CXIP4 in Arabidopsis development and survival, as well as its involvement in pre-mRNA splicing. Results CXIP4 is an essential gene, required for plant survival and proper development Most ZCCHC-containing proteins studied to date participate in RNA metabolism (Aceituno-Valenzuela et al. 2020). Arabidopsis CXIP4 (AT2G28910) is a single-copy gene with two exons (Fig. 1A), the second of which encodes a protein of 332 amino acids (aa) containing a ZCCHC motif (residues 81 to 98; Supplementary Fig. S1). As annotated in the UniProtKB database (https://www.uniprot.org/uniprotkb), CXIP4 also contains two intrinsically disordered regions (IDRs) flanking the ZCCHC motif which occupy a large part of the protein (residues 33 to 59 and 124 to 332); the latter IDR being part of a region rich in arginine (R; residues 198 to 289; Supplementary Fig. S1). Structure of the CXIP4 gene and developmental phenotypes of cxip4 insertional alleles. A) Schematic representation of the CXIP4 gene, indicating the positions of its start (ATG) and stop (TAG) codons and the T-DNA insertions (triangles) in the cxip4 alleles studied in this work. Boxes represent exons, with untranslated (UTRs) and coding regions shown in white and black, respectively. B–M) Morphological vegetative phenotypes of (B, E, H) Col-0, (C, F, I) cxip4-1, (D, G, J) cxip4-2, (K) cxip4-1/cxip4-2, (L) cxip4-1 CXIP4pro:CXIP4:GFP, and (M) cxip4-2 CXIP4pro:CXIP4:GFP plants. Photographs were taken (B–D) 7, (E–G) 14, and (H–M) 21 days after stratification (das). Scale bars: (B, C, D, F, G, I, and K) 1 mm, and (E, H, J, L, and M) 5 mm. To date, only a single functional analysis of the CXIP4 gene in Arabidopsis has been performed, which was based on transgenic lines with β-estradiol-inducible overexpression of the full-length cDNA (Coego et al. 2014). To investigate the effects of the loss of function of CXIP4, we obtained two T-DNA insertional lines with disruptions in the coding region of AT2G28910: GK-537C02 and SALK_044245 (Fig. 1A). GK-537C02 seeds either did not germinate or produced phenotypically wild-type plants or callus-like seedlings with early arrested development (Fig. 1, B, C, E, F, H, and I). The phenotypically wild-type plants were found to be homozygous for the wild-type allele of CXIP4 or hemizygous for the T-DNA insertion; siliques of the latter plants exhibited some light-green, immature seeds that contained early arrested embryos (Fig. 2, A to D). We sowed 3,626 additional seeds from hemizygous plants, finding that 15.6% exhibited a wrinkled appearance and failed to germinate (12.6%) or produced dark callus-like seedlings (3%), while the remaining 84.4% gave rise to phenotypically wild-type plants (Fig. 2, E and F). We genotyped 30 callus-like seedlings and 170 phenotypically wild-type plants. All aberrant seedlings were homozygous for the T-DNA insertion, whereas those with a wild-type phenotype were hemizygous or homozygous for the wild-type allele of CXIP4. We named the insertional allele of CXIP4 carried by the GK-537C02 line cxip4-1. The cxip4-1/cxip4-1 seedlings did not develop true leaves or any other recognizable organ (Fig. 1, C, F, and I). The ungerminated seeds were assumed to also be homozygous for the cxip4-1 allele, and we did not genotype them. Developmental defects caused by the cxip4-1 allele. A, B) Dissected immature siliques from (A) Col-0 and (B) CXIP4/cxip4-1 plants; the latter exhibit light-green seeds (black arrows). C, D) Embryos in (C) dark- and (B) light-green seeds from siliques of CXIP4/cxip4-1 plants. E, F) Seeds from (E) Col-0 and (F) CXIP4/cxip4-1 plants; the latter developed some wrinkled seeds (black arrows), whose genotype was found to be cxip4-1/cxip4-1. Photographs in (A–D) were taken 60 das. Scale bars: (A, B, E, and F) 1 mm, and (C, D) 100 µm. Some seeds of the SALK_044245 line yielded plants with slowed growth and defects in both vegetative and reproductive development (Fig. 1, D, G, and J; Fig. 3; Supplementary Fig. S2). Seven days after stratification (das), plants homozygous for the T-DNA insertion were smaller than Col-0 plants, and some had three cotyledons (8.8% of 979 seedlings; Fig. 1, B and D). 14 to 21 das, the mutant plants showed pointed and reticulated leaves with a reduced density of trichomes, which varied considerably among leaves of different individuals, and whose branching was not affected; rosette leaves also produced high levels of anthocyanins (Fig. 1, E, G, H, and J; Fig. 3, A to J). These plants flowered late, developed numerous shoots with reduced apical dominance, and remained green for more than 80 days (Supplementary Fig. S2). In addition, their flowers produced short siliques containing likely undeveloped ovules, along with a few mature seeds that were larger than those of Col-0 (Fig. 3, K to O). Pleiotropy of the morphological phenotype of cxip4-2 plants. A–C) Trichome density and branching on the third leaves of the wild-type Col-0 and the cxip4-2 mutant. A) Total number of trichomes. B) Trichome density. C) Percentage of trichomes with different levels of branching. The trichomes of ten leaves per genotype were collected 21 (Col-0) or 29 (cxip4-2) das. D–J) Anthocyanin accumulation in the leaves of cxip4-2. D–G) Adaxial and abaxial surfaces of rosettes of Col-0 and cxip4-2 collected 12 das. H, I) Anthocyanin extracts from (H) Col-0 and (I) cxip4-2. J) Anthocyanin contents of ten rosettes per genotype, each collected 12 das. K, L) The (K) silique length, and (L) fertility of Col-0 and cxip4-2 plants. Ten siliques collected from five plants per genotype were analyzed. M, N) Dissected immature siliques from (M) Col-0 and (N) cxip4-2 plants, with the latter exhibiting undeveloped ovules (black arrows). O) Sizes of Col-0 and cxip4-2 seeds, determined from 60 seeds per genotype. Error bars in (A–C), (J–L), and (O) represent standard deviations. Asterisks in (A), (J–L), and (O) indicate values significantly different from the wild type, determined using Student's t-test (*P < 0.0001). Scale bars: (D–G) 1 mm and (M, N) 0.5 mm. The SALK_044245 line contains two additional annotated T-DNA insertions besides the one disrupting the CXIP4 gene: one in AT2G23640 and another downstream of the 3′ untranslated region (3′-UTR) of AT3G50030 (Supplementary Fig. S3). We did not eliminate these insertions because (i) the ones in AT2G23640 and CXIP4 (AT2G28910) are linked, being 2.36 megabases (Mb) apart. The other T-DNA insertion is located in an intergenic region, 88 bp downstream of AT3G50030, as we reconfirmed by Sanger sequencing, in which we also identified a duplicated T-DNA 58 bp upstream of the annotated one. (ii) Both genes, AT2G23640 and AT3G50030, are not expressed in the wild-type plants, except AT2G23640 that expresses in seeds, as described in eFP Browser (Klepikova et al. 2016). (iii) We obtained T-DNA insertional lines disrupting the coding regions of AT2G23640 and AT3G50030 genes, and confirmed that these lines do not exhibit a mutant phenotype in any vegetative or reproductive developmental step (Supplementary Fig. S3). We named the insertional allele of CXIP4 carried by the SALK_044245 line cxip4-2. The T-DNA insertions in cxip4-1 and cxip4-2 are at similar locations (Fig. 1A), but their mutant phenotypes are quite different (Fig. 1, C, D, F, G, I, and J). Therefore, to further examine if cxip4-1 and cxip4-2 are null, we used several combinations of primers, flanking the insertions or hybridizing within the T-DNA and the CXIP4 gene (Supplementary Fig. S4A and Table S1), to perform reverse transcription PCR (RT-PCR) analyses. We detected transcripts upstream of the insertions in both cxip4-1 and cxip4-2, although the levels were low in cxip4-2. As expected, we did not detect transcripts using primers flanking the insertions, and only cxip4-2 showed slight transcription downstream of the insertion (Supplementary Fig. S4B). We then sequenced the PCR products obtained using primers that hybridize within the T-DNA and the CXIP4 gene, which resulted in chimeric sequences combining CXIP4 and T-DNA (Supplementary Fig. S4C). The mRNA from cxip4-1 was predicted to encode a mutant CXIP4-1 protein four aa longer than the wild-type CXIP4, with only 271 aa shared between the mutant and wild type. Translation of the chimeric cxip4-2 mRNA would result in a protein of 240 aa missing part of the C-terminal region (Supplementary Figs. S1 and S4C). The aberrant CXIP4-1 protein appears to lack activity, but the truncated CXIP4-2 protein may be partially functional. These results might explain the stronger phenotype observed in cxip4-1 plants compared with cxip4-2. Since cxip4-1 presents a much stronger phenotype than cxip4-2, and considering the results obtained through RT-PCR, we concluded that cxip4-1 is likely to be a null allele of CXIP4 whereas cxip4-2 is hypomorphic. We carried out a complementation test by crossing CXIP4/cxip4-1 to cxip4-2/cxip4-2 plants. We obtained four cxip4-1/cxip4-2 F1 plants from two different crosses, one of which displayed three cotyledons, as did some cxip4-2/cxip4-2 plants. The four cxip4-1/cxip4-2 heterozygous plants displayed a strong mutant phenotype, more similar to that of cxip4-1/cxip4-1 than to cxip4-2/cxip4-2 homozygotes, and their development was arrested after the emergence of cotyledons and one or two small leaves (Fig. 1K). These results confirm the allelism of cxip4-1 and cxip4-2, and with the recessive nature of both, suggest that they are null and hypomorphic alleles of CXIP4, respectively. To confirm that the disruption of CXIP4 by the T-DNA insertions in cxip4-1 and cxip4-2 was the cause of their mutant phenotypes, we transferred a copy of the wild-type CXIP4 gene fused to green fluorescent protein (GFP) into CXIP4/cxip4-1 and cxip4-2/cxip4-2 plants. Indeed, homozygous cxip4-1 CXIP4pro:CXIP4:GFP and cxip4-2 CXIP4pro:CXIP4:GFP plants were phenotypically wild type (Fig. 1, L and M), confirming in the last plants that vegetative and reproductive mutant traits of cxip4-2 (growth retardation, leaf and rosette morphology and size, trichome density, anthocyanin levels, flowering time, plant architecture, silique development, and size and number of seeds; Figs. 1 and 3; Supplementary Fig. S2) were normalized. CXIP4 is ubiquitously expressed in Arabidopsis Cis-regulatory elements in Arabidopsis are usually located within the 500-bp region upstream of the transcription start site (TSS), with the majority of these elements (86%) located proximal to the TSS (+1 bp position), spanning from −1,000 bp to +200 bp, and a low density of regulatory sequences at more than 1.5 kb from the TSS (Yu et al. 2016). AT2G28910 (CXIP4) and AT2G28920 are oriented head-to-head, with their respective 5′ ends toward each other and their translation start codons 1,026 bp away from each other, sharing an intergenic region of 401 bp that could potentially contain a bidirectional promoter. No 5′-UTR is annotated for AT2G28920, but three different splice variants have been found for CXIP4, differing only in the length of the 5′-UTR, due to the AS of its single intron that interrupts this region. Indeed, the coding region of CXIP4 begins in its second exon, and we supposed that the first exon and its single intron could harbor regulatory sequences. Based on this information, we constructed two transgenes by cloning the 2,104- and 971-bp genomic regions upstream of the translation start codon of CXIP4 into the pMDC164 vector, including 1,479 (CXIP4proI) or 346 (CXIP4proII) bp from the TSS of CXIP4 and that in the CXIP4proI included the entire AT2G28920 gene. The sizes of these regions were also conditioned by the presence of several stretches of repetitive nucleotides, which were avoided in the primer design. These promoters were fused to the β-glucuronidase (GUS) gene to drive its expression. In the CXIP4proI:GUS and CXIP4proII:GUS transgenic plants, the highest GUS activity was detected in the emerging leaves and the vasculature of the cotyledons, leaves, and roots, especially in the root apex, but it was much lower in flowers (Supplementary Fig. S5, A to D), which is consistent with results previously described using an RNA gel blot analysis of RNA from the leaves, roots, stems, and flowers (Cheng et al. 2004). We also detected GUS activity during embryogenesis in seeds at the green mature embryo stage, which is consistent with the early lethality found in cxip4-1 (Supplementary Fig. S5E). We did not observe any differences between plants expressing the CXIP4proI:GUS or CXIP4proII:GUS transgenes, and the expression of the CXIP4proII:CXIP4:GFP transgene rescued all traits of the mutant phenotypes of cxip4-1 and cxip4-2 plants (Fig. 1, L and M), suggesting that the 971-bp region upstream of the translation start codon of CXIP4 contains all the regulatory elements needed for CXIP4 expression. Phylogenetic analysis of CXIP4 and other PTHR31437 family members CXIP4 appears to be plant-specific, although it is classified as a member of the SREK1IP1 protein family (PTHR31437; Supplementary Table S2) according to the PANTHER database (https://www.pantherdb.org; Thomas et al. 2022). Human SREK1IP1 and Arabidopsis CXIP4 share only a 21-aa region, which harbors the 18-aa ZCCHC motif (Supplementary Fig. S6). It is of note that the ZCCHC motifs of CXIP4 and SREK1IP1 clearly grouped in a branch of the unrooted tree that we obtained from the alignment of 198 ZCCHCs from 110 proteins of yeast, human, and Arabidopsis (Aceituno-Valenzuela et al. 2020). This grouping reflects the remarkably high similarity (13 identical aa) of the ZCCHC motifs of these two proteins (Supplementary Fig. S6). The PTHR31437 family, as described in the PANTHER database, comprises 96 members from the animal (SREK1IP1-like), plant (CXIP4-like), protozoa, and chromista kingdoms, with proteins of this family expected to be present in all organisms of these four kingdoms. While all animal and most plant genomes studied to date encode only one SREK1IP1/CXIP4 protein (such as Arabidopsis), some plants contain two to four co-orthologs. This is the case in wheat, whose hexaploid genome encodes four CXIP4 proteins, three of which are almost identical and originate from its three subgenomes (Supplementary Table S2). We studied the amino acid sequences of the 96 proteins classified as members of the PTHR31437 family in the UniProtKB database. One of the two proteins of the ancestral angiosperm Amborella trichopoda and those from the chromistan Phytophthora ramorum and Thalassiosira pseudonana lacked the ZCCHC motif (Supplementary Table S2). We aligned the sequences of five ZCCHC-containing proteins from five distant organisms: Chlamydomonas reinhardtii (UniProtKB accession number A0A2K3DE08), Arabidopsis thaliana (Q84Y18), Physcomitrium (Physcomitrella) patens (A0A2K1K6W5), Drosophila melanogaster (Q9W3Z5), and Homo sapiens (Q8N9Q2). The alignment revealed high conservation only in the region containing the ZCCHC motif and in three isolated basic residues of K or R, located downstream of the ZCCHC motif and at the C-terminus of the proteins (Supplementary Fig. S6). We found five additional conserved amino acids, in addition to the three canonical C and one H of the ZCXCHC motif, which may be relevant for the functions of PTHR31437 family proteins. The alignment also showed that the Drosophila melanogaster and human proteins lack the N-terminal extension found in the CXIP4-like proteins of photosynthetic organisms. In addition, the ZCCHC motifs in Drosophila melanogaster and humans are located at the N-terminal regions of these proteins (Supplementary Fig. S6), as also detected in all metazoan SREK1IP1 orthologs, whose lengths were half those of plant CXIP4-like proteins (Supplementary Table S2). We also aligned the sequences of putative CXIP4-like proteins from different angiosperm lineages, selecting species encoding a single ortholog: Hordeum vulgare, Arabidopsis thaliana, Capsicum annuum, Vitis vinifera, and Populus trichocarpa. This second alignment revealed higher conservation than that shown in Supplementary Fig. S6, as expected, including a 48-aa region containing the ZCCHC motif (Supplementary Fig. S7) and the full conservation of the N-terminal 39 aa, which are absent in animals and partially conserved in the single-celled green alga Chlamydomonas reinhardtii (Supplementary Figs. S6 and S7). In fact, when we performed a BLASTP search using that 39-aa conserved sequence as a query, we only found the exact sequence in CXIP4 proteins from divergent land plant lineages of five major clades: bryophytes (such as Physcomitrium patens), lycophytes, monilophytes, gymnosperms, and angiosperms (such as Arabidopsis). CXIP4 is a nucleoplasmic protein When transiently expressed in tobacco (Nicotiana tabacum) plants, Arabidopsis CXIP4 localized to the nucleus in a diffuse pattern; however, when expressed in yeast and tobacco BY-2 cells, it was also found in the cytoplasm, forming discrete spots that did not correspond to mitochondria (Cheng et al. 2004). More recently, Arabidopsis CXIP4 was found to exclusively localize to the nuclei of chickpea (Cicer arietinum) protoplasts (Cheng and Nakata 2020). As an initial assessment of the localization of Arabidopsis CXIP4, we used the MULocDeep web server (https://www.mu-loc.org/; Jiang et al. 2021, 2023), which predicts the localization of any eukaryotic protein in 44 suborganellar compartments based on its primary sequence. We included human SREK1IP1 in the analysis. MULocDeep predicted that both proteins localize to the nucleus, with scores of 0.840 and 0.996, respectively (Supplementary Fig. S8A). Within the nucleus, both proteins are predicted to predominantly localize to the nucleoplasm, nucleolus, and nuclear speckles (Supplementary Fig. S8B), which are known to be enriched in pre-mRNA splicing factors, snRNAs, and polyadenylated [poly(A)+] RNAs (Belmont 2022). Using the LOCALIZER (https://localizer.csiro.au/; Sperschneider et al. 2017) and NoD (https://www.compbio.dundee.ac.uk/www-nod/index.jsp; Scott et al. 2010, 2011) web tools, we predicted nuclear and nucleolar localization signals (NLSs and NoLSs, respectively) in the C-terminal half of CXIP4. These NLSs and NoLSs signals were identified in two or three separate stretches, respectively, each composed of positively charged aa, mainly K and R (Supplementary Fig. S1). To determine the localization of Arabidopsis CXIP4, we analyzed Col-0 plants carrying the CXIP4pro:CXIP4:GFP transgene, which successfully rescued the mutant phenotypes of cxip4-1 and cxip4-2 plants (Fig. 1, L and M), revealing that the CXIP4-GFP fusion protein is functionally equivalent to the endogenous CXIP4. The CXIP4-GFP fusion protein appeared in a speckled pattern within the nucleus (Fig. 4), which may correspond to nuclear speckles, as predicted by MULocDeep (Supplementary Fig. S8B). Subcellular localization of CXIP4 in Arabidopsis roots collected five das. A–F) Confocal laser-scanning micrographs of Col-0 plants homozygous for the CXIP4pro:CXIP4:GFP transgene. Fluorescent signals correspond to (A, D) 4′,6-diamidino-2-phenylindole (DAPI) nuclear staining (in blue), (B, E) CXIP4-GFP (in green), and (C, F) their overlap. Scale bars: 100 μm. Very recently, Arabidopsis CXIP4 and its orthologs in wheat, tomato, and rice have been detected in multiple nuclear foci during a transient expression assay in wheat protoplasts (He et al. 2024). These foci appear very similar to those found in CXIP4pro:CXIP4:GFP plants in a Col-0 background. cxip4-2 globally affects the expression of genes involved in pathogen defense To gain insight into the biological processes affected by the loss of function of CXIP4, we performed RNA-seq analysis using poly(A)+ RNAs from Col-0 and cxip4-2 plants collected 14 das. We identified 2,160 deregulated genes in cxip4-2 compared to the wild type, with a similar number of up- (1,094) and downregulated (1,066) genes (Fig. 5A; Supplementary Data Set 1). CXIP4 was the tenth most downregulated gene, suggesting that the T-DNA insertion disrupting CXIP4 in the cxip4-2 mutant strongly reduces its transcript levels (Fig. 5, A and B; Supplementary Data Set 1), as observed in the abovementioned RT-PCR analysis (Supplementary Fig. S4B). However, compared with the early lethality exhibited by cxip4-1, the low expression of CXIP4 in cxip4-2 suggests that it is sufficient for viability, but strongly compromises fertility, in addition to causing a highly pleiotropic developmental phenotype (Figs. 1, B to J, 2 and 3; Supplementary Fig. S2). Differential gene expression analysis between Col-0 and cxip4-2 plants. A) Volcano plot of differentially expressed genes in cxip4-2 compared to Col-0 plants, collected 14 das. Upregulated genes (fold change ≥ 2) and downregulated genes (fold change ≤ 0.5) and with FDR < 0.05 (Benjamini–Hochberg false discovery rate) are shown as yellow and cyan dots, respectively; nonderegulated genes appear as gray dots. Horizontal and vertical red dashed lines mark the cutoffs of the negative decimal logarithm of FDR and binary logarithm of fold changes. The dot marked with a red arrow represents CXIP4, which was downregulated in cxip4-2, as expected. B) Plot of aligned reads for CXIP4 in Col-0 and cxip4-2, obtained with Integrative Genomics Viewer (IGV) software (http://software.broadinstitute.org/software/igv/). Gene structure is represented as described in Fig. 1 legend. C) Bar plot representation of the overrepresented Gene Ontology (GO) Slim terms for deregulated genes in cxip4-2. These terms were identified through statistical overrepresentation tests using PANTHER (https://pantherdb.org/) GO Slim annotation sets for biological processes, molecular functions, and cellular components (fold enrichment > 1 and FDR < 0.05). Redundant terms, with an identical set of genes to others that were already included, are not plotted. For a separate analysis of upregulated and downregulated genes in cxip4-2, see Supplementary Data Set 1. Gene Ontology (GO) enrichment analysis of the deregulated genes in cxip4-2 plants revealed an overrepresentation of terms primarily associated with pathogen defense (Fig. 5C). In a separate analysis of downregulated genes, we also found an overrepresentation of terms related to indole glucosinolate, camalexin, and phytoalexin metabolism, as well as salicylic acid synthesis and perception, among others. By contrast, upregulated genes were mainly associated with responses to sulfur starvation and other stress conditions, as well as flavonoid, glucosinolate, and jasmonic acid metabolism (Supplementary Data Set 1). Taken together, these results suggest that the deregulation of CXIP4 causes a global defense response to biotic and abiotic stress. Interestingly, the upregulated genes were also enriched in DNA-binding transcription factor activity, including 80 out of 1,208 genes in the reference database, mainly with the C2H2 zinc finger (eight genes) and homeobox (32 genes) DNA-binding domains (Supplementary Data Set 1). Several upregulated genes in cxip4-2 may explain its mutant phenotype The cxip4-2 mutant is viable but exhibits a highly pleiotropic phenotype, suggesting that many developmental pathways or key genes are deregulated in the mutant. Therefore, we inspected the RNA-seq data, looking for specific deregulated genes that may explain some of the phenotypes exhibited by cxip4-2 plants. Interestingly, GLABROUS1 (GL1), encoding a positive regulator of trichome development (Larkin et al. 1994), was downregulated (Supplementary Data Set 1), which could explain the low density of trichomes found in cxip4-2 leaves, whereas trichome branching, which is not controlled by this gene, remained unaffected (Fig. 3, A to C). Among the upregulated genes in cxip4-2, we also identified ENHANCED RNA INTERFERENCE-1-LIKE-1 (ERIL1; Supplementary Data Set 1), which encodes a ribonuclease H-like protein involved in the processing of chloroplast pre-rRNAs (Mermigka et al. 2016). Transgenic plants overexpressing ERIL1 are defective in chloroplast rRNA maturation, resulting in the accumulation of several rRNA biogenesis intermediates. These precursors include a 1.7-kb transcript that may correspond to the 16S rRNA containing the intergenic region that separates this gene from the trnl gene, as well as the 23S-4.5S dicistronic precursor (3.2 kb), and the 2.4 and 2.9 kb incompletely processed 23S rRNAs (Mermigka et al. 2016; Supplementary Fig. S9A). The upregulation of ERIL1 in cxip4-2 could explain the presence of two additional peaks in the profiles of total RNA samples, which we obtained before their use in the RNA-seq assay. Both peaks are located between those corresponding to the cytosolic 18S and 25S rRNAs, whose sizes in Arabidopsis are approximately 1.8 and 3.4 kb, respectively, and could correspond to the incompletely processed chloroplast 23S rRNAs mentioned above (Supplementary Fig. S9). PRODUCTION OF ANTHOCYANIN PIGMENT 1 (PAP1) was also found among the upregulated genes in cxip4-2 (Supplementary Data Set 1). PAP1 is a transcription factor that directly induces the transcription of genes involved in anthocyanin biosynthesis. PAP1-overexpressing plants accumulate high amounts of anthocyanins when grown in the presence of high sucrose concentrations, which cause stress (Teng et al. 2005). The upregulation of PAP1, along with the enriched GO terms “anthocyanin-containing compound biosynthetic process” and “flavonoid metabolic process” (Supplementary Data Set 1), may also account for the high anthocyanin levels observed in the cxip4-2 mutant (Fig. 3, D to J), suggesting that cxip4-2 plants are constitutively stressed. We validate by RT-qPCR the deregulation of GL1, ERIL1, PAP1, and CXIP4, using cDNA from Col-0, cxip4-2, and cxip4-2 CXIP4pro:CXIP4:GFP plants (Supplementary Fig. S10). We included LIPOXYGENASE 2 (LOX2) in the validation as one of the most upregulated genes in cxip4-2 plants belonging to one of the most enriched GO terms: “jasmonic acid metabolic process” (Supplementary Data Set 1 and Fig. S10). We did not find enriched GO terms related to the UPR, as might have been anticipated given the resistance to DTT and TM observed in plants overexpressing CXIP4 (Hossain et al. 2016). Nevertheless, among the downregulated genes in cxip4-2 plants was AT1G42990, which encodes the BASIC REGION/LEUCINE ZIPPER MOTIF 60 (bZIP60) transcription factor. bZIP60 activates the transcription of key genes involved in UPR signaling (Iwata and Koizumi 2005). Pre-mRNA splicing is defective in cxip4-2 plants As previously mentioned, the GO annotations of human SREK1IP1 and its interactions associate this protein with pre-mRNA splicing. Thus, we analyzed the effects of cxip4-2 on pre-mRNA splicing, finding 939 differential AS events in cxip4-2 plants compared with Col-0. The most frequent were intron retentions (IR), with 543 events (57.8% of the total differential AS events), mainly resulting in the retention of the alternatively spliced intron (Fig. 6A; Supplementary Data Set 2). Analysis of differential AS between Col-0 and cxip4-2 plants. A) Summary of differential AS events and misspliced genes identified in cxip4-2 compared to Col-0 18 das. Schematic representations depict five different types of AS events, with black or gray rectangles and lines representing constitutive or alternatively spliced exons and introns, respectively, and red lines marking the splice junctions. B) Bar plot of overrepresented GO terms and PANTHER protein classes for the misspliced genes in cxip4-2. These terms were identified as described in Fig. 5 legend, but using GO Complete and PANTHER protein class annotation sets. Only the most specific hierarchically organized GO terms and PANTHER protein classes were plotted; for a complete list, see Supplementary Data Set 2. The second most frequent differential AS events identified in cxip4-2 plants compared with Col-0 were alternative 3′ splice site (A3′SS) events, totaling 255 (27.2%). Using Integrative Genomics Viewer (IGV; Robinson et al. 2011), we determined that 85 out of these 255 events (33.3% of the differential A3′SS events) involved tandem 3′SSs (acceptor sites) with NAGNAG sequences (Supplementary Data Set 2), in which A3′SSs were exactly three nucleotides apart from each other (in-frame); their alternative usage would result in proteins differing by one aa, which might not affect their function. The tandem 3′SS motif NAGNAG is overrepresented in genes encoding RNA-recognition motifs (RRMs), including SR proteins (Schindler et al. 2008). However, we did not find enrichment in any protein class for these 85 A3′SS events, which included genes related to pre-mRNA splicing. Subsets of misspliced genes in cxip4-2 are related to different steps of mRNA metabolism We analyzed the overrepresented GO terms for misspliced genes in cxip4-2, finding that they were mainly related to the regulation of cuticular wax biosynthesis, RNA-mediated gene silencing, mRNA splicing, chromatin remodeling, and DNA repair, among others (Fig. 6B; Supplementary Data Set 2). Five out of 15 genes in the reference database were categorized under the “ta-siRNA processing” term: SUPPRESSOR OF GENE SILENCING 1 (SGS1; also known as ANAC052), SGS2 (also known as RNA DEPENDENT RNA POLYMERASE 6 [RDR6]), and SGS3, with differential IR events; and DICER-LIKE 3 (DCL3) and DOUBLE-STRANDED-RNA-BINDING PROTEIN 4 (DRB4), with differential A3′SS events. These five genes, whose expression was not deregulated in cxip4-2 plants, were also classified under the term “siRNA processing”, which also included the nonderegulated genes NUCLEAR RNA POLYMERASE D1B (NRPD1B) and TOUGH (TGH), exhibiting differential IR and A3′SS events, respectively (Supplementary Data Set 2). Other overrepresented GO terms with several overlapping genes included “post-transcriptional gene silencing”, “regulatory ncRNA processing”, “RNA-mediated gene silencing”, and “negative regulation of gene expression, epigenetic” (Supplementary Data Set 2). We also analyzed the PANTHER protein classes that were overrepresented among the misspliced genes in cxip4-2, which included the “mRNA polyadenylation factor” class, with six out of 39 genes in the reference database. These six genes encode NOT1 (the scaffold protein of the CCR4-NOT complex), the deadenylases CATABOLITE REPRESSOR 4C (CCR4C) and CCR4D, POLY(A) POLYMERASE 2 (PAPS2) and PAPS4 (also known as NPAP), and MATERNAL EFFECT EMBRYO ARREST 44 (MEE44). The differential AS events included IR events in CCR4D, PAPS2, PAPS4, and MEE44, and A3′SS events in NOT1 and CCR4C, although these genes were not deregulated in cxip4-2. These proteins are thought to function in polyadenylation-assisted RNA degradation through the exonucleolytic cleavage of the poly(A) tail, thereby contributing to mRNA export, nuclear quality control, and translation (Passmore and Coller 2022). We aimed to validate some of these events by RT-PCR. We ruled out validating A5′SS and A3′SS events because, in most cases, the length difference when using the canonical SS was only a few nucleotides. Additionally, we excluded genes that exhibited combinations of different AS events at the same coordinates, as we assumed they would be difficult to analyze. Consequently, we selected five genes from different functional categories, each of which had only one single significant intron retention event detected. These genes were SGS1 (ANAC052) with 30.5% of IR relative to Col-0; SGS3, with 21.6%; NRPD1B, with 47.93%, representing the second highest percentage of IR from a gene with a single AS event detected; CCR4D, with 10.07%; and MEE44, with 21.77% of IR. We also included AT2G01100 in the validation, with 47.6% of IR relative to Col-0 (Fig. 7, A to G; Supplementary Data Set 2). Validation of differential intron retention (IR) events in the cxip4 mutants and accumulation of poly(A)+ RNAs in the nuclei of cxip4-2 leaf cells. A–G) RT-PCR analysis of six differentially spliced introns in the Col-0, cxip4-1, cxip4-2, and cxip4-2 CXIP4pro:CXIP4:GFP (rescue) plants. Schematic representations illustrate the structures of the (A) SGS1, (B) SGS3, (C) NRPD1B, (D) CCR4D, (E) MEE44, (F) AT2G01100, and (G) ACT2 genes. Numbers indicate the positions in the pre-mRNA of the flanking exons corresponding to the analyzed introns (shown in gray) affected by the alternative IR events, with red lines marking the splice junctions. Arrows represent the primers used (not drawn to scale) and are labeled as follows: F1 (corresponding to SGS1-F in Supplementary Table S1), R1 (SGS1-R), F2 (SGS3-F), R2 (SGS3-R), F3 (NRPD1B-F), R3 (NRPD1B-R), F4 (CCR4D-F), R4 (CCR4D-R), F5 (MEE44-F), R5 (MEE44-R), F6 (AT2G01100-F), R6 (AT2G01100-R), F7 (ACT2-F), and R7 (ACT2-R). Genomic DNA (gDNA) from Col-0 and water were used as positive and negative controls for the PCR amplifications, respectively, while the expression of the housekeeping gene ACT2 acted as an internal control to normalize the cDNA quantity used. H–M) Poly(A)+ RNA-FISH assays in palisade mesophyll cells of the first-node leaves of (H–J) Col-0 and (K–M) cxip4-2. Fluorescent signals correspond to (H, K) DAPI nuclear staining, (I, L) fluorescein from an oligo-dT probe targeting poly(A)+ RNAs, and (J, M) their merged images. Confocal laser-scanning micrographs were taken from five leaves per genotype of plants collected 14 das for Col-0 and 22 das for cxip4-2. The gain of the 515/30 nm detector was kept constant to allow for direct comparisons of fluorescein intensity between samples. Scale bars: 50 µm. We initially used cDNAs from three biological replicates of Col-0, cxip4-2, and cxip4-2 CXIP4pro:CXIP4:GFP. However, changes in the relative mRNA isoform proportions compared to the wild type were not evident for some of these genes. Subsequently, we repeated the RT-PCR experiments, including a sample of cDNA from cxip4-1 escaper plants, which we isolated from several sowings, as we had observed that the number of these plants decreased in successive generations of CXIP4/cxip4-1 heterozygous individuals. The results obtained with cxip4-1 cDNA were highly noticeable (Fig. 7, A to G), confirming the involvement of CXIP4 in pre-mRNA splicing and highlighting the different functional nature of the cxip4 alleles, with cxip4-2 being hypomorphic and cxip4-1 likely a null allele. cxip4-2 plants accumulate poly(A)+ RNAs Eukaryotic cells prevent the translation of pre-mRNAs by different mechanisms. For example, pre-mRNAs can be retained in the nucleus until they are fully or correctly spliced (Palazzo and Lee 2018; Wegener and Müller-McNicoll 2018; Rudzka et al. 2022). This mechanism also ensures efficient pre-mRNA splicing by factors that are confined to the nucleus. In addition, mRNA export factors are preferentially recruited to spliced mRNAs (reviewed in Reed 2003; Kelly and Corbett 2009), and pre-mRNAs or misspliced RNAs that still leak into the cytoplasm are degraded by nonsense-mediated mRNA decay (NMD; reviewed in Bhuvanagiri et al. 2010). The prp8-7 hypomorphic allele of PRE-MRNA PROCESSING 8 (PRP8), encoding a central factor of the spliceosome, causes global missplicing, and we previously detected the nuclear accumulation of poly(A)+ RNAs in prp8-7 plants (Cabezas-Fuster et al. 2022). We aimed to determine whether cxip4-2 plants accumulate mRNAs in the nucleus and we performed poly(A)+ RNA fluorescence in situ hybridization (RNA-FISH) assays using a fluorescein-labeled oligo(dT) probe, detecting poly(A)+ RNAs accumulated to high levels within the nuclei of cxip4-2 leaf palisade mesophyll cells (Fig. 7, H to M). Accumulation of poly(A)+ RNAs in the nucleus of cxip4-2 can be due to the presence of misspliced mRNAs that are retained. Discussion CXIP4 is an essential gene whose partial loss of function triggers pronounced pleiotropic changes No mutational analyses of CXIP4-like genes have been described, and only a single functional analysis of the CXIP4 gene in Arabidopsis has been performed, which used transgenic lines with β-estradiol-inducible overexpression of full-length cDNAs (Coego et al. 2014). Subsequent analysis of these lines revealed the resistance of CXIP4 overexpressing plants to DTT and TM, suggesting a role for CXIP4 in the UPR (Hossain et al. 2016). However, gene overexpression can cause artifacts which do not occur in loss-of-function analysis. Moreover, loss-of-function analysis is usually more informative in the absence of functional redundancy. This would be the case for CXIP4, since it is a single-copy gene in Arabidopsis (Supplementary Table S2). The functional analysis of two T-DNA insertional lines of CXIP4 harboring alleles with different degrees of loss of function allowed us to conclude that CXIP4 is an essential gene for embryogenesis but is also required for postembryonic development (Figs. 1 to 3; Supplementary Fig. S2). cxip4-1 is likely a null allele of CXIP4 that mainly causes early embryonic lethality, and the few embryos that escape this lethality produce callus-like plants with arrested development (Fig. 1, A, B, C, E, F, H, I, and L; Fig. 2). The cxip4-2 allele, characterized by significantly reduced CXIP4 expression, as identified by RNA-seq and RT-qPCR analyses (Fig. 5, A and B; Supplementary Fig. S10), is fully viable but causes a striking pleiotropic phenotype when homozygous. This phenotype includes severe growth retardation leading to delayed flowering, abnormal cotyledon and leaf morphology characterized by sparse trichome density, the loss of apical dominance, and markedly reduced fertility (Fig. 1, A, B, D, E, G, H, J, and M; Fig. 3; Supplementary Fig. S2). Our analysis revealed the deregulation of several key genes, predominantly encoding transcription factors, in cxip4-2 plants, providing insights into the underlying mechanisms of its pleiotropic phenotype (Supplementary Data Set 1 and Fig. S10). Furthermore, our findings suggest that the C-terminal region of CXIP4, which is absent from the CXIP4-2 mutant protein, is dispensable for plant survival but crucial for proper development (Supplementary Fig. S1). The pleiotropy exhibited by cxip4-2 and the embryonic lethality of cxip4-1 match the broad GUS activity detected in almost all tissues, including embryos, where expression was very high (Supplementary Fig. S5). Besides their aberrant morphological phenotype, cxip4-2 plants accumulate high levels of anthocyanins, suggesting that they are constitutively stressed. This is consistent with the finding that the deregulated genes identified by RNA-seq analysis were enriched in GO terms involved in global defense responses (Fig. 3, D to J; Fig. 5C), and with the elevated overexpression of PAP1 (Supplementary Data Set 1 and Fig. S10), encoding a well-known transcription factor that regulates anthocyanin biosynthesis (Teng et al. 2005). These results also agree with the recently proposed function of TaCAXIP4, one of the four putative co-orthologs of CXIP4 in wheat (Supplementary Table S2), in the calcium-mediated plant immune response against pathogens. This role of TaCAXIP4 results from its interaction with TaHRC, a histidine-rich calcium-binding nuclear protein conferring susceptibility to Fusarium head blight or scab, a destructive disease in wheat worldwide caused by Fusarium graminearum (Su et al. 2019). It has been proposed that the interaction between TaHRC and TaCAXIP4 leads to the hijacking of TaCAXIP4, resulting in the suppression of calcium-mediated immune responses (Chen et al. 2022). Arabidopsis CXIP4 and human SREK1IP1 are highly divergent orthologs The PANTHER database contains the complete proteomes of 143 organisms, including model organisms and others used in biomedical and biotechnological research. A search in PANTHER revealed that metazoan SREK1IP1-like and plant CXIP4-like proteins form the PTHR31437 family, but their sequences are highly divergent. In fact, they are classified as LDOs (least diverged orthologs), which are the most nearly “equivalent” gene pairs between different organisms based on phylogenetic analysis. Metazoan SREK1IP1-like proteins are approximately half the length of plant CXIP4-like proteins, sharing only some aa from the ZCCHC motif (Supplementary Table S2 and Fig. S6). However, it is of note that the ZCCHC motifs of Arabidopsis CXIP4 and human SREK1IP1, which share 13 to the 18 aa, clearly grouped in the same branch of an unrooted tree generated by phylogenetic analysis based on sequence alignment of 198 ZCCHC motifs from 110 ZCCHC-containing proteins in yeast (7), Arabidopsis (69), and humans (34) that we identified in a systematic search (Aceituno-Valenzuela et al. 2020). This grouping was likely due to the high number of conserved amino acids, in addition to the three canonical cysteines (C) and one histidine (H) that define the ZCCHC motif. Despite the low similarity between Arabidopsis CXIP4 and human SREK1IP1 outside the ZCCHC domain, their subcellular and suborganellar localizations appear to be highly similar, as predicted by MULocDeep (Supplementary Fig. S8). MULocDeep uses a machine learning method based on primary sequence information (https://www.mu-loc.org/; Jiang et al. 2021, 2023) to identify common signals or patterns in two proteins. In fact, the NLS and NoLS signals were present in the C-terminal half of CXIP4, which is rich in basic aa, similar to human SREK1IP1. These findings support the notion that these proteins share similar functions. The specific categories of misspliced genes in cxip4-2, rather than their number, may explain its mutant phenotype The prp8-7 hypomorphic allele of PRP8, encoding a central factor of the spliceosome, causes global missplicing, as revealed by its 8,124 increased IR events compared to Col-0 (Sasaki et al. 2015), which is 18 times higher than in the cxip4-2 plants (446 increased IR events; Fig. 6A; Supplementary Data Set 2). However, the morphological phenotype of cxip4-2 is much more severe than that of prp8-7 with both mutants exhibiting a partial loss of function of the corresponding essential genes (CXIP4 and PRP8). Therefore, the phenotype of cxip4-2 plants is not caused by the number of missplicing events found in the RNA-seq analysis, which suggested a minor role for CXIP4 in general splicing. However, RT-PCR validation of six misspliced genes showing IR events, including cDNA from cxip4-1 escaper plants, revealed a more prominent role of CXIP4 in pre-mRNA splicing, as the ratio of mRNAs with retained introns to fully spliced mRNAs was much higher in cxip4-1 compared with cxip4-2 plants (Fig. 7, A to G). We also found that most of the detected differential AS events do not deregulate the expression of the affected genes, as is the case with the prp8a-14 mutant of the PRP8 gene (Llinas et al. 2022). It is possible that the missplicing events that we found in cxip4-2 plants cause a reduction of functional mRNAs and alter the relative concentrations of the resulting protein isoforms, which may have diverse functional consequences. However, the overrepresentation of specific GO terms, mainly “gene silencing by DNA methylation” or the protein class “mRNA polyadenylation factor” among the misspliced genes in cxip4-2 plants (Supplementary Data Set 2) also suggests a role for CXIP4 in transcriptional and post-transcriptional regulation of gene expression, which should be studied further. CXIP4 could play a key role in the modulation of the AS In GeneCards in the Human Gene Database (https://www.genecards.org/), SREK1IP1 is described as having GO annotations related to “Possible splicing regulator involved in the control of cellular survival”. We found relevant information on human SREK1IP1 in the HitPredict database (http://www.hitpredict.org), which compiles physical protein–protein interactions from six different databases and 124 species, all experimentally identified in both small-scale and high-throughput assays (López et al. 2015). An interaction with NKAP obtained through an affinity purification mass spectrometry (AP-MS) high-throughput assay (Huttlin et al. 2017) is described in HitPredict, along with several interactions with splicing factors (Supplementary Table S3). These results confirm the conservation of interactions between SREK1IP1/CXIP4 and NKAP/MAS2 in humans and Arabidopsis, supporting a role for human SREK1IP1 in pre-mRNA splicing. Additionally, TaCAXIP4, one of the four CXIP4-like co-orthologs in wheat, has been proposed to be part of a nuclear speckled complex with FHB, based on Y2H assays using FHB as bait. This complex is thought to consist of these two proteins along with others whose orthologs participate in AS regulation, including RNA-BINDING PROTEIN LUC7-LIKE and two SR-rich proteins (He et al. 2024). The Arabidopsis CXIP4 could also play a key role in the formation of nuclear complexes that regulate AS and/or mRNA export. We observed that most of the differentially spliced introns identified in our RNA-seq analysis also show some retention in Col-0, with higher levels in the cxip4-2 mutant (Supplementary Data Set 2), and even more pronounced retention in cxip4-1. However, we can only confirm the latter for the six IR events analyzed by RT-PCR (Fig. 7, A to G). These results suggest that the 5′SSs are weak in these misspliced introns, and that CXIP4 facilitates their recognition. This hypothesis aligns with the proposed composition of the complex in wheat, based on interactions with HRC, as previously mentioned, where CXIP4 was identified alongside LUC7 (He et al. 2024). In Arabidopsis, LUC7 is encoded by three genes, whose products are involved in both general and AS. Similar to cxip4-2 plants, RNA-seq analysis of the Arabidopsis luc7 triple mutant revealed only 640 differential splicing events compared with the wild type, with only 17 of these events (2.7%) affecting transcript levels. Additionally, in the luc7 triple mutant, the export of mRNAs with retained introns to the cytoplasm is blocked, suggesting that these transcripts evade the NMD mechanism, which normally operates in the cytoplasm (de Francisco Amorim et al. 2018). The nuclear retention of misspliced mRNAs in the luc7 triple mutant is consistent with our RNA-FISH assay (Fig. 7, H to M), where we found an accumulation of poly(A)+ RNAs in the nuclei of cxip4-2 leaf cells, likely corresponding to misspliced mRNAs. Furthermore, in yeast, Luc7p is involved in 5′SS recognition, and in luc7 loss-of-function mutants, splicing of introns with nonconsensus 5′SS or branchpoint sequences is more defective than in the wild type (Fortes et al. 1999). Since SR proteins act as positive regulators of AS at weak splice sites, CXIP4 may collaborate with them in splice site recognition. Our conclusions regarding the function of Arabidopsis CXIP4 are based on the developmental and molecular phenotypes associated with its cxip4-1 and cxip4-2 loss-of-function alleles. However, the predicted CXIP4.1 and CXIP4.2 mutant proteins retain the ZCCHC motif, which is the sole motif identified in CXIP4-like and SREK1IP1-like proteins and represents their primary conserved feature. This motif is commonly found in proteins involved in mRNA and noncoding RNA metabolism, mediating interactions with other proteins and RNAs (reviewed in Aceituno-Valenzuela et al. 2020). Furthermore, the ZCCHC motif is the fourth most common RNA-binding motif in human proteins, highlighting its functional importance, and supporting its classification as an RNA-binding motif (Ray et al. 2023). Generating Arabidopsis CXIP4 mutants specifically lacking the ZCCHC motif, or introducing similar mutations in other CXIP4-like or SREK1IP1-like proteins, would be invaluable for clarifying the functional significance of this motif and for understanding its evolutionary conservation across the plant and animal kingdoms. Materials and methods Plant material, growth conditions, and genotyping All Arabidopsis thaliana (L) Heynh. lines used in this work were in the Columbia-0 (Col-0) genetic background. The Col-0 wild type and the T-DNA insertional lines GABI_537C02 (cxip4-1), SALK_044245 (cxip4-2), SALK_097293, and SALK_017259 (Alonso et al. 2003; Kleinboelting et al. 2012) were obtained from the Nottingham Arabidopsis Stock Center (NASC; Nottingham, United Kingdom) and propagated in our laboratory for further analysis. Seed sterilization and sowing, plant culture, and crosses were performed as previously described (Ponce et al. 1998; Berná et al. 1999), except that plant agar was replaced with 6 g·L−1 of Gelrite (Duchefa Biochemie). When required, culture media were supplemented with hygromycin (15 µg·mL−1) and kanamycin (50 µg·mL−1). To genotype the T-DNA insertional lines, genomic DNA was extracted from the samples as described in Ponce et al. (2006), and the presence of the T-DNA insertions was verified by PCR amplification using the primers shown in Supplementary Table S1. Most Sanger sequencing reactions and electrophoreses were carried out in our laboratory with ABI PRISM BigDye Terminator Cycle Sequencing kits and an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems). Some sequencing reactions were carried out at Stab Vida (Caparica, Portugal). RT-PCR and RT-qPCR analysis Total RNA was isolated with TRIzol Reagent (Invitrogen) from three biological replicates per genotype, each consisting of 100 (Col-0, cxip4-2, and cxip4-2 CXIP4pro:CXIP4:GFP) or 20 (cxip4-1) mg of plants collected 14 or 21 das. Prior to cDNA synthesis, the RNA was treated with TURBO DNase (Invitrogen). RT-PCR amplifications were performed as described in Casanova-Sáez et al. (2014). The mRNAs produced by the cxip4-1 and cxip4-2 mutant alleles were analyzed by Sanger sequencing of cDNAs. qPCR amplifications were carried out in a Step-One Real-Time PCR System (Applied Biosystems) using three technical replicates per biological replicate. The primers used are shown in Supplementary Table S1. The housekeeping gene ACTIN2 (ACT2) served as an internal control for normalization and relative quantification of gene expression. The CT values obtained were normalized using the 2−ΔΔCT method (Livak and Schmittgen 2001). Statistical analyses of the ΔCT values between Col-0 and cxip4-2, and between Col-0 and cxip4-2 CXIP4pro:CXIP4:GFP, were performed using unpaired Student's t-tests with the GraphPad t-test web calculator (https://www.graphpad.com/quickcalcs/ttest1/). Construction of transgenes and analysis of transgenic lines Transgenes were generated by Gateway cloning as described in Sánchez-García et al. (2015), using the pGEM-T Easy221 entry vector (provided by B. Scheres) and the pMDC107 and pMDC164 (Curtis and Grossniklaus 2003) destination vectors. To assess the temporal and spatial expression patterns of CXIP4, two constructs, CXIP4proI:GUS and CXIP4proII:GUS, were generated by PCR amplification of the 2,104- and 971-bp genomic regions upstream of the translation start codon of CXIP4. The PCR products were subcloned into the pMDC164 destination vector. GUS enzymatic activity was observed throughout the development of Col-0 plants homozygous for the CXIP4proI:GUS and CXIP4proII:GUS transgenes. All parts of the plants were imaged using bright-field microscopy under a Nikon D-Eclipse C1 laser-scanning confocal microscope. GUS staining was performed as previously described (Donnelly et al. 1999). The samples were incubated in X-Gluc buffer overnight at 37 °C. The seed coat and embryo were separated before incubating the seeds in GUS staining solution. To determine the subcellular localization of CXIP4, the CXIP4proII:CXIP4:GFP construct was generated by PCR amplification of the 971-bp genomic region upstream of the translation start codon and the full-length coding sequence of CXIP4 (without its stop codon), which was then subcloned into the pMDC107 destination vector. The structural integrity of all constructs was verified by sequencing prior to their transfer into Arabidopsis plants by Agrobacterium tumefaciens-mediated transformation via the floral dip method (Clough and Bent 1998). Primers used to obtain these constructs and for Sanger sequencing are described in Supplementary Table S1. Analysis of plant morphology Rosette photographs were taken under a Nikon SMZ1500 stereomicroscope equipped with a Nikon DXM1200F or DS-Ri2 digital camera. For large rosettes, high-resolution images were obtained by taking multiple pictures of the same plant and assembling them with the Photomerge tool of Adobe Photoshop CS3 software. Adult plants were photographed with a Canon PowerShot SX200 IS camera. Trichome density and branch number were determined using the third leaves of ten Col-0 and cxip4-2 plants collected 21 and 29 das, respectively. The leaves were cleared with ethanol and chloral hydrate, mounted on slides, and photographed under a Leica DMRB microscope equipped with a Nikon DXM1200 digital camera (Nadi et al. 2023). The number of trichomes with two to five branches was scored for each leaf and expressed as a percentage of the total number of trichomes. For the analysis of siliques and seeds, photographs were taken under a Nikon SMZ1500 stereomicroscope equipped with either a Nikon DXM1200F or DS-Ri2 digital camera. Silique length and number of seeds per silique were measured from ten siliques collected from five plants per genotype. Leaf and seed silhouettes (ten and 60 per genotype, respectively) were drawn on a Cintiq 18SX Interactive Pen Display (Wacom) using Adobe Photoshop CS3, and their areas were measured with the NIS Elements AR 3.1 image analysis software (Nikon). Statistical analyses of the trichome density and branching, silique length, number of seeds per silique, and seeds size in Col-0 and cxip4-2 were performed using unpaired Student's t-tests with the GraphPad t-test web calculator. Anthocyanin extraction and measurement To quantify anthocyanin contents, ten biological replicates per genotype were used, each consisting of a rosette collected 12 das and weighed. Anthocyanins were extracted and homogenized in 45% (v/v) methanol and 5% (v/v) acetic acid buffer using glass beads and a MixerMill 400 (Retsch) automatic mixer. The samples were centrifuged twice for 5 min at 13,500 × g to remove cellular debris. The absorbance of each clarified extract was measured spectrophotometrically at 530 and 657 nm for anthocyanin and contaminating chlorophyl in acidic extraction buffer, respectively (Mancinelli 1990). Anthocyanin content was calculated as A530 − (0.25 · A657) · g−1 of rosette fresh weight, where 25% of the A657 reading was subtracted to account for chlorophyl degradation products. A statistical analysis of the anthocyanin content in Col-0 and cxip4-2 plants was performed using unpaired Student's t-test with the GraphPad t-test web calculator. RNA-seq and analysis of AS Total RNA was extracted using TRIzol Reagent (Invitrogen) from three biological replicates per genotype, each consisting of 40 to 80 mg of rosettes collected 14 das. At least 6 µg of RNA was sent to Novogene (Cambridge, United Kingdom) for sequencing and subsequent gene expression profiling. An Agilent 2100 bioanalyzer with an RNA 6000 Nano Kit (Agilent Technologies) was used to assess RNA concentration and integrity. Total RNA samples underwent mRNA enrichment prior to random fragmentation. Libraries were produced with a NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs) and sequenced on an Illumina NovaSeq 6000 platform using a 2 × 150 bp run protocol. More than 50 million nonstranded 150 bp paired-end reads were generated from each library. All FASTQ files were submitted to the Sequence Read Archive (SRA) database of the National Center for Biotechnology Information (NCBI) under the BioProject accession number PRJNA1090680 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1090680). Differential gene expression analysis was conducted by Novogene. Briefly, raw data were processed through in-house Perl scripts to remove low-quality reads and reads containing adapters or more than 10% uncertain nucleotides. Clean reads were aligned to the Arabidopsis Col-0 genome (TAIR10) with HISAT2 2.0.5 (Kim et al. 2019) and assembled with StringTie 1.3.3b (Pertea et al. 2015), using default parameters. The number of reads (counts) mapped to each gene was determined using FeatureCounts 1.5.0-p3 (Liao et al. 2014) with default parameters, giving the number of fragments per kilobase of transcript per million mapped reads (FPKM). The identification of differentially expressed genes between Col-0 and cxip4-2 plants was performed using the DESeq2 R package 1.20.0, with the selection criteria set at a fold change ≥ 2 or ≤ 0.5, for up- or downregulated genes, and a Benjamini–Hochberg false discovery rate (FDR) < 0.05. Differential AS analysis was carried out at the Bioinformatics for Genomics and Proteomics Unit of the Centro Nacional de Biotecnología (CNB, Madrid). Briefly, the quality and purity of the raw reads were determined using FastQC 0.11.9 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and FastQ Screen 0.14.1 (Wingett and Andrews 2018), respectively. Reads were aligned to the Arabidopsis Col-0 genome (TAIR10) using STAR 2.7.10a (Dobin et al. 2013) with default parameters. Differential AS events between Col-0 and cxip4-2 plants were assessed with rMATS-turbo 4.1.2 (Shen et al. 2014). Only those events with an FDR < 0.05, an absolute Delta (percent spliced-in, PSI) > 0.1 (10%), and an average number of reads covering the event > 5 were considered to be statistically significant. Statistical analyses for the overrepresented GO terms and PANTHER protein classes among the differentially expressed or spliced genes between Col-0 and cxip4-2 plants were performed using Fisher's Exact tests with FDR correction, using the GO database released 2023-01-05 (DOI: 10.5281/zenodo.7942786) and PANTHER 18.0 (released 2023-08-01; https://pantherdb.org/; Thomas et al. 2022). The selection criteria were set at a fold enrichment > 1 and an FDR < 0.05. RNA-FISH and confocal microscopy Poly(A)+ RNA-FISH assays were carried out as previously described (Parry et al. 2006; Micol-Ponce et al. 2020), using first-node leaves from Col-0 and cxip4-2 plants collected 14 or 22 das, respectively, and a 40-mer fluorescein-labeled oligo(dT) probe (synthesized by Eurofins Genomics) at a concentration of 0.5 μg·mL−1 in PerfectHyb Plus Hybridization Buffer (Sigma-Aldrich). To visualize fluorescent signals from the RNA-FISH assays and from the CXIP4:GFP fusion protein, samples were mounted on slides with Vectashield antifade mounting medium (Vector Laboratories) or water containing 0.5 µg·mL−1 of 4′,6-diamidino-2-phenylindole (DAPI). All images were taken under a Nikon D-Eclipse C1 confocal microscope with Nikon EZ-C1 operation software, or a Leica Stellaris eight STED microscope with Leica Application Suite X (LAS X) software. DAPI, fluorescein, and GFP were excited at 408, 488, and 543 nm and their emissions detected at 450/35 nm, 515/30 nm, and 650 LP nm, respectively. Bioinformatic analysis using protein sequences The multiple sequence alignments shown in Supplementary Figs. S6 and S7 were obtained using MUSCLE V3.8 (Edgar 2004) from the EMBL-EBI bioinformatic web server (Madeira et al. 2022), with default parameters. To assess the subcellular and suborganellar localizations of the Arabidopsis CXIP4 protein, the MULocDeep web server (https://www.mu-loc.org/; Jiang et al. 2021, 2023) was utilized. To predict the effect of eliminating part of the C-terminal half of CXIP4-2, the LOCALIZER (https://localizer.csiro.au/; Sperschneider et al. 2017) and NoD (https://www.compbio.dundee.ac.uk/www-nod/index.jsp; Scott et al. 2010, 2011) web tools were utilized. Accession numbers Sequence data from this article can be found at The Arabidopsis Information Resource (TAIR; https://www.arabidopsis.org) under the following accession numbers: CXIP4 (AT2G28910), AT2G23640, AT3G50030, GL1 (AT3G27920), ERIL1 (AT3G15140), PAP1 (AT1G56650), LOX2 (AT3G45140), SGS1 (AT3G10490), SGS3 (AT5G23570), NRPD1B (AT2G40030), CCR4D (AT1G31500), MEE44 (AT4G00060), AT2G01100, and ACT2 (AT3G18780). Supplementary Material Acknowledgments The authors would like to thank J.A. García-Martín for the differential AS analyses; J. Castelló, D. Navarro, and M. Gomariz for their excellent technical assistance; and J.L. Micol for useful discussions and comments on the manuscript, as well as for the use of his facilities. Contributor Information Uri Aceituno-Valenzuela, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. Sara Fontcuberta-Cervera, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. Rosa Micol-Ponce, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. Raquel Sarmiento-Mañús, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. Alejandro Ruiz-Bayón, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. María Rosa Ponce, Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain. Author contributions M.R.P. obtained funding and conceived, designed, and supervised research. R.M.-P and R.S.-M. partially co-supervised this work. U.A.-V., S.F.-C., R.M.-P., R.S.-M., and A.R.-B. conducted the experiments. U.A.-V., S.F.-C., R.M.-P., and M.R.P. analyzed the data, created the figures, tables, and datasets and wrote the manuscript. All authors revised and approved the manuscript. Supplementary data The following materials are available in the online version of this article. Supplementary Figure S1. Localization of the ZCCHC motif, R-rich region, and nuclear and nucleolar localization signals in the predicted wild-type CXIP4 and mutant CXIP4-1 and CXIP4-2 proteins. Supplementary Figure S2. Reproductive development of Col-0 and cxip4-2 plants over time. Supplementary Figure S3. Structure of the AT2G23640 and AT3G50030 genes and developmental phenotypes of individuals from the SALK_097293 and SALK_017259 lines. Supplementary Figure S4. Molecular effects of the cxip4 mutations on the expression of the CXIP4 gene. Supplementary Figure S5. Spatial expression of CXIP4 during Arabidopsis development. Supplementary Figure S6. Sequence conservation among putative eukaryotic CXIP4 orthologs. Supplementary Figure S7. Sequence conservation among putative CXIP4 orthologs in angiosperms. Supplementary Figure S8. Predicted localizations of Arabidopsis CXIP4 and human SREK1P1 proteins. Supplementary Figure S9. Defects in 23S rRNA maturation in cxip4-2 plants. Supplementary Figure S10. RT-qPCR analysis of the relative expression of CXIP4, GL1, ERIL1, PAP1, and LOX2 in Col-0, cxip4-2, and cxip4-2 CXIP4pro:CXIP4:GFP plants. Supplementary Table S1. Oligonucleotides used in this work. Supplementary Table S2. Summary of protein-coding genes assigned to the PTHR31437 family according to the PANTHER database. Supplementary Table S3. Summary of high-confidence physical protein–protein interactions for human SREK1IP1, identified from high-throughput experiments according to the HitPredict database. Supplementary Data Set 1. Deregulated genes in cxip4-2 plants. Supplementary Data Set 2. Alternative splicing events in cxip4-2 plants. Funding This work was supported by grants from the Ministerio de Ciencia, Innovación y Universidades of Spain (PID2020-117125RB-I00 [MCI/AEI/FEDER, UE]) and the Generalitat Valenciana (PROMETEO CIPROM/2022/2) to M.R.P. R.M.-P. held a María Zambrano distinguished researcher contract funded by the Next-Generation EU programs and administered by the Universidad Miguel Hernández. U.I.A.-V. held a predoctoral fellowship from the Generalitat Valenciana (GRISOLIAP/2016/134). Data availability The data underlying this article are available in the article and in its online supplementary material except the FASTQ files that were submitted to the Sequence Read Archive (SRA) database of the National Center for Biotechnology Information (NCBI) under the BioProject accession number PRJNA1090680 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1090680). Dive Curated Terms The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper: CAX1 Gramene: AT2G38170 CAX1 Gramene: AT2G38170 CAX1 Araport: AT2G38170 CAX1 Araport: AT2G38170 NRPD1b Gramene: A08g503970.1_BraROA NRPD1b Gramene: A08g503970.1_BraROA NRPD1b Araport: A08g503970.1_BraROA NRPD1b Araport: A08g503970.1_BraROA CXIP4 Gramene: AT2G28910 CXIP4 Gramene: AT2G28910 CXIP4 Araport: AT2G28910 CXIP4 Araport: AT2G28910 DCL3 Gramene: AT3G43920 DCL3 Gramene: AT3G43920 DCL3 Araport: AT3G43920 DCL3 Araport: AT3G43920 DRB4 Gramene: AT3G62800 DRB4 Gramene: AT3G62800 DRB4 Araport: AT3G62800 DRB4 Araport: AT3G62800 NPAP Gramene: AT4G32850 NPAP Gramene: AT4G32850 NPAP Araport: AT4G32850 NPAP Araport: AT4G32850 DAPI CHEBI: CHEBI:51231 DAPI CHEBI: CHEBI:51231 References Associated Data Supplementary Materials Data Availability Statement The data underlying this article are available in the article and in its online supplementary material except the FASTQ files that were submitted to the Sequence Read Archive (SRA) database of the National Center for Biotechnology Information (NCBI) under the BioProject accession number PRJNA1090680 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1090680).