Functional and evolutionary analysis of the Arabidopsis 4R-MYB protein SNAPc4 as part of the SNAP complex

Abstract

Transcription initiation of the genes coding for small nuclear RNA (snRNA) has been extensively analyzed in humans and fruit fly, but only a single ortholog of a snRNA-activating protein complex (SNAPc) subunit has so far been characterized in plants. The genome of the model plant Arabidopsis thaliana encodes orthologs of all three core SNAPc subunits, including A. thaliana SNAP complex 4 (AtSNAPc4)—a 4R-MYB-type protein with four-and-a-half adjacent MYB repeat units. We report the conserved role of AtSNAPc4 as subunit of a protein complex involved in snRNA gene transcription and present genetic evidence that AtSNAPc4 is an essential gene in gametophyte and zygote development. We present experimental evidence that the three A. thaliana SNAPc subunits assemble into a SNAP complex and demonstrate the binding of AtSNAPc4 to snRNA promoters. In addition, co-localization studies show a link between AtSNAPc4 accumulation and Cajal bodies, known to aggregate at snRNA gene loci in humans. Moreover, we show the strong evolutionary conservation of single-copy 4R-MYB/SNAPc4 genes in a broad range of eukaryotes and present additional shared protein features besides the MYB domain, suggesting a conservation of the snRNA transcription initiation machinery along the course of the eukaryotic evolution.

The snRNA-activating protein complex in plants, described here functionally, consists of three evolutionary conserved subunits and is essential for gametophyte and zygote development.

Introduction

Small nuclear RNAs (snRNAs) are a class of short, noncoding, nonpolyadenylated transcripts with distinct secondary structures that are highly conserved throughout eukaryotes. Formation of specific secondary structures is essential for the association of the snRNAs with certain sets of protein factors and thus ensures the creation of small ribonucleic particles (snRNPs). These snRNPs are the main components of the spliceosome (summarized in Matera and Wang, 2014). The snRNPs U1, U2, U4, U5, and U6 are designated according to the type of uridine-rich (U) snRNA included in the respective RNA/protein particle. Together with other spliceosomal proteins, the various snRNPs catalyze the splicing of canonical GU-AG introns out of precursor messenger RNA (Mount, 1982; Burset et al., 2000).

Although dissimilar in their sequence, the snRNA gene promoters of Homo sapiens (Hs), Drosophila melanogaster (Dm), and Arabidopsis thaliana (At) all contain an important cis-element positioned 50- to 70-bp upstream of the transcription start site. Depending on the model system, the elements have been designated as proximal sequence element (PSE; Hs), proximal sequence element A (Dm: PSEA), or upstream sequence element (USE; At), respectively (Dahlberg and Lund, 1988; Vankan and Filipowicz, 1989; Zamrod et al., 1993). The factor binding to this promoter has been best described in H. sapiens. Here, the PSE is recognized by a multi protein complex termed snRNA-activating protein complex (SNAPc; Sadowski et al., 1993), which is synonymously known as PSE-binding protein (PBP; Waldschmidt et al., 1991) and PSE-binding transcription factor (PTF; Murphy et al., 1992). Binding of HsSNAPc to the PSE is crucial for the recruitment of RNA polymerases (Pol) II and III to the respective snRNA gene promoters (Dergai et al., 2018), and thus activates the promoter and causes transcription of snRNA genes. HsSNAPc is composed of five subunits: HsSNAPc1 (SNAP43, PTF γ), HsSNAPc2 (SNAP45, PTF δ), HsSNAPc3 (SNAP50, PTF β), HsSNAPc4 (SNAP190, PTF α), and HsSNAPc5 (SNAP19; Henry et al., 1995, 1996, 1998; Yoon et al., 1995; Bai et al., 1996; Sadowski et al., 1996; Yoon and Roeder, 1996; Wong et al., 1998).

It has been shown that the so-called “mini-SNAPc,” consisting of the three core subunits HsSNAPc1, HsSNAPc3, and the amino-terminal third of HsSNAPc4, is capable of binding specifically to the PSE and directing snRNA gene transcription in vitro (Mittal et al., 1999). For D. melanogaster, a PSEA-binding protein complex (DmPBP) that activates snRNA gene transcription has been identified as well (Su et al., 1997). Since the subunits of this complex have been shown to be orthologs of the core HsSNAPc subunits, it has been renamed DmSNAPc (Li et al., 2004). In the model flowering plant A. thaliana, the SHOOT REDIFFERENTIATION DEFECTIVE 2 (SRD2, At1g28560) gene, which encodes an ortholog of HsSNAPc3, was characterized as an activator of USE-dependent snRNA transcription (Ohtani and Sugiyama, 2005). Further components of the A. thaliana SNAP complex (AtSNAPc) have not yet been characterized.

The first MYB protein identified is encoded by the oncogene v-Myb that is present in the genome of the avian myeloblastosis virus (Klempnauer et al., 1982). The MYB domain, which is the shared characteristic of members of the MYB protein family, generally comprises one or several imperfect repeats, each forming a helix-turn-helix structure of ∼52 amino acid (aa). Often, this domain is involved in protein/DNA recognition (Jia et al., 2004). The MYB domain-containing subunit HsSNAPc4 was shown to be responsible for sequence-specific binding of HsSNAPc to the PSE, which is further stabilized by DNA binding functions of the HsSNAPc1 subunit (Wong et al., 1998; Jawdekar et al., 2006). The composition of the MYB domain of HsSNAPc4 and the D. melanogaster homolog DmSNAPc4, with four (and a half) adjacent MYB repeats, characterizes the proteins as members of the 4R-MYB class of the MYB protein family (Li et al., 2004). In the first sequenced plant genome, that of A. thaliana (The Arabidopsis Genome Initiative, 2000), a single gene (AtMYB4R1, At3g18100) was identified to encode an MYB domain protein with a corresponding number of MYB repeats, which showed the 4R-MYB class to also be present in plants (Stracke et al., 2001). Other higher plant genomes (e.g. Vitis vinifera, Oryza sativa, Physcomitrella patens, and Beta vulgaris) were also shown to contain a single-copy gene encoding a class 4R-MYB protein (Yanhui et al., 2006; Matus et al., 2008; Dubos et al., 2010; Stracke et al., 2014).

There are several indications that the SNAP complex associates with Cajal bodies (CBs), usually roughly spherical mobile sub-organelles found in the nucleus of eukaryotic cells. CBs occur in varying size and number, depending on the proliferative and metabolic state of the cell (reviewed in Cioce and Lamond, 2005). The DNA-encoding snRNA genes are often found adjacent to CBs (Frey and Matera, 1995; Smith et al., 1995) and subunit SNAPc2 was identified to be enriched in CBs (Schul et al., 1998). Moreover, CBs are prominent in cells showing high transcriptional activity and thus high spliceosomal activity which results in increased snRNA demand (Ogg and Lamond, 2002). This fits to snRNA gene transcription occurring in association with CBs (Jacobs et al., 1999; Frey and Matera, 2001; Dundr et al., 2007).

This study reports the molecular identification and characterization of the A. thaliana gene MYB4R1/SNAPc4 (At3g18100), which encodes a protein with an MYB domain consisting of four and a half adjacent MYB repeats. This 4R-MYB domain is homolog to the MYB domains of HsSNAPc4 and DmSNAPc4. We report the evolutionary conservation of 4R-MYB proteins in a broad range of eukaryotes and present evidence that AtSNAPc4 is a part of the AtSNAP complex which is found in the nucleus. This is the first experimental description of the AtSNAPc.

Results

So far, little attention has been paid to 4R-MYB proteins, particularly in plants. This eukaryotic protein class was first described by Wong et al. (1998) who reported that the HsSNAPc4 protein contains four MYB repeats. The first mentioning of a plant gene encoding a 4R-MYB protein was in a study that described the MYB gene family of A. thaliana (Stracke et al., 2001). A later overview of MYB transcription factors by Dubos et al. (2010) reported that 4R-MYB proteins are present as single-copy genes in a small number of sequenced plant genomes.

4R-MYB genes are conserved in a comprehensive selection of eukaryotic genomes

In order to discover distinguishing features of the 4R-MYB protein class, we performed a survey of a broad range of 100 eukaryotic genome sequences, representing major clades of the eukaryotic evolutionary tree of life, for the presence of 4R-MYB genes. The aa sequence of the MYB domain of the AtSNAPc4 protein was used as query against protein sequence databases. Selection of true orthologs to AtSNAPc4 was performed by manual curation of BLASTP search results, based on similarity values and conservation of the regular spaced bulky aromatic residues characteristic for MYB repeats (Ogata et al., 1992). Our survey revealed a single-copy 4R-MYB gene in most of the analyzed genomes. This also holds for the majority of the analyzed land plant (embryophyta) genome sequences. The orthologs identified for H. sapiens and D. melanogaster are the previously characterized SNAPc4 subunits of HsSNAPc and DmSNAPc (Wong et al., 1998; Li et al., 2004), which are also encoded as single-copy genes in the respective genomes.

Two copies were identified in 12 of the 60 analyzed embryophyte genomes, and three copies in the genome of Arabidopsis lyrata (Supplemental Figure S1). A closer examination of the gene loci encoding the three A. lyrata 4R-MYB proteins (Supplemental Figure S2) showed that AtSNAPc4/At3g18100 is very similar to Al3g31260 (∼90%) and Al7g46060 (∼86%). The third A. lyrata 4R-MYB locus Al7g34390 might be originating from a short segmental duplication together with a gene fusion, since the region upstream does not align with the At3g18100 locus and the last two exons align to a locus on A. thaliana chromosome 4 (At4g19630).

In a more detailed comparison of the 4R-MYB protein sequences, we observed between 40% and 96% aa sequence similarity of the various 4R-MYB domains to that of AtSNAPc4. The length of the deduced polypeptides and positioning of the MYB domain within the protein turned out to be somehow diverse within the SNAPc4 sequences (Figure 1). For the identification of evolutionarily conserved aa residues of the MYB domain of 4R-MYBs, a multiple sequence alignment (MSA) of the protein sequences was produced and analyzed. In order to prevent highly similar sequences distorting the identification of residues conserved throughout all eukaryotes, more redundant sequences of closely related species were removed from the sequence set (Neuwald et al., 1995) resulting in a subset of protein sequences from 23 species. The MSA of these AtSNAPc4 orthologs revealed a high sequence diversity in the whole-length proteins. The alignment of the four MYB repeat sequences, designated R_a, R_b, R_c, and R_d, according to the nomenclature used for the HsSNAPc4 MYB domain (Wong et al., 1998), shows a high conservation of aa residues known to be of structural and functional importance in MYB repeats (Supplemental Figure S3B). The additional half MYB repeat (R_h), located N-terminal to R_a (Wong et al., 1998), is also conserved in the AtSNAPc4 orthologs.

Figure 1.

4R-MYB genes in eukaryotic genomes. Left, guide tree with arbitrary branch lengths, based on the phylogenetic framework as presented by Du et al. (2015). Only a fraction of the plant genomes included in the survey is shown. Middle, percentage of AAs with similar chemical properties, determined by pairwise alignments of the 4R-MYB protein sequences to the MYB domain of AtSNAPc4. Right, schematic representations that show the positioning of the 4R-MYB domain and other conserved features in the protein sequences. The genome of A. coerulea (marked by an asterisk) contains two 4R-MYB genes, the displayed similarity value and protein structure represent the protein encoded by the gene Aqcoe3G054500.

A number of protein features have been described for the two previously characterized animal 4R-MYB proteins HsSNAPc4 and DmSNAPc4, but besides the MYB domain, none of them are shared. Hence, we analyzed our compiled ortholog set for additional conserved elements of the 4R-MYB protein class. Using a MEME motif search, we identified a 28 aa residues long motif located ∼150–200 aa N-terminal of the MYB domain in 93% of the SNAPc4 orthologs (Supplemental Figure S3, A and C). The most conserved residues of the motif are separated alternately by two and three less conserved aa residues. Given the repetitive seven residues structure, we refer to this conserved signature as SNAPc4 heptad motif. In AtSNAPc4, the motif (aa 238–258) is localized in a region predicted to form an α-helix (aa 224–269) as secondary structure. As α-helices display circa 3.6 aa per turn (Chothia et al., 1981; as indicated by the sine wave in Supplemental Figure S3C), the most conserved residues would reside on the same side of the helix. N-terminal to the SNAPc4 heptad motif, we noted the repeated occurrence of D/E-rich regions (≥7 of 10 residues being aspartic or glutamic acid, as defined by Chou and Wang, 2015), predominantly in 4R-MYB proteins of land plants and algae. A more detailed analysis, with a lowered threshold of six D/E residues, revealed that almost all of the examined land plant protein N-termini contain up to four regularly spaced D/E-rich regions within N-terminal 135 aa (Figure 1). In the protein sequence of AtSNAPc4 we identified four D/E-rich regions, the third repeat possibly originating from a local duplication of the second (aa 61–96; 99–131; Supplemental Figure S3A). An analysis of the aa composition of all N-terminal sequences of the orthologs (upstream of the SNAPc4 heptad motif) revealed a general enrichment in negatively charged residues (12.8% D, 10.1% E) as well as serine (11.0% S) when compared to the protein sequences downstream of the SNAPc4 heptad motif (5.6% D, 7.7% E, and 8.5% S) or eukaryotic nonmembrane proteins in general (Gaur, 2014). The conserved sequence features in the 4R-MYB proteins might indicate a conservation of protein function.

AtSNAPc4 splice variants

The AtSNAPc4 protein is encoded by the gene locus At3g18100, for which four differently spliced transcripts (AtSNAPc4.1–AtSNAPc4.4) were described in the A. thaliana Araport11 annotation (Cheng et al., 2017). Based on this clear indication of alternative splicing, we aimed to identify all alternatively spliced transcripts for AtSNAPc4 by analyses of cDNA constructs. We identified a total of six splice variants (Figure 2), of which the two most frequently occurring were the previously known variants AtSNAPc4.1 and AtSNAPc4.2. Newly identified splice variants are AtSNAPc4.5 and AtSNAPc4.6. We also found splice variant AtSNAPc4.4 but did not identify a cDNA sequence corresponding to splice variant AtSNAPc4.3. Furthermore, we were unable to find experimental evidence for the existence of splice variant AtSNAPc4.3, marked by a discrete 65-nt-long intron, in reads from publicly available RNA-Seq data sets from various developmental stages and tissues. Except for AtSNAPc4.1, all established transcripts result in a frameshift which introduces a premature termination codon (PTC) 726 nucleotides (variants 2 and 6) or 519 nucleotides (variants 4 and 5) downstream of the start codon in relation to variant AtSNAPc4.1. The mature mRNA of isoform 1 is created by removal of six introns from the primary transcript (seven exons are kept, see Figure 2). The corresponding coding sequence (CDS) encodes an 847 aa long protein with the MYB domain located from aa residue 415–645. If translated, the resulting peptides of splice variants 2, 4, 5, and 6 would not contain the MYB domain. Since splice variant AtSNAPc4.1 is the only experimentally confirmed one encoding a 4R-MYB protein, we used this transcript variant for our further analyses.

Figure 2.

Schematic representation of the AtSNAPc4 splice variants. The striped boxes indicate the MYB domain coding region. The first five splice variants represent the variants we identified in our study, named according to the annotation for the locus At3g18100 in the Araport11 annotation. Splice variants 5 and 6 were newly identified in this study, based on cDNA sequencing. No experimental evidence was found establishing the existence of splice variant AtSNAPc4.3, which is depictured in gray. The asterisk marks the variant-defining intron.

AtSNAPc4 expression pattern

To get an initial overview of AtSNAPc4 expression patterns, we analyzed RNA-Seq data mapped on the Araport11 genome annotation (https://jbrowse.arabidopsis.org/index.html?data=Araport11&loc=Chr3%3A6200312..6204962&tracks=Araport11_Loci%2CAraport11_gene_models&highlight=). A weak expression of AtSNAPc4 is evident in all of the available datasets (light- and dark-grown seedling, root, aerial, leaf, root and shoot apical meristem, Stage 12 flower, receptacle, carpel, and pollen), with the highest read coverage of the At3g18100 locus in the light-grown seedlings and the leaf dataset. Also when using the TRAVA RNA-Seq database (http://travadb.org/), low AtSNAPc4 expression was found in all 78 tissues analyzed, with enhanced read coverage in dry seeds. For a more detailed examination of AtSNAPc4 expression, we generated stable transgenic A. thaliana lines carrying a proAtSNAPc4:GUS construct, with a 1-kb AtSNAPc4 promoter fragment controlling the expression of the reporter uidA [encoding β-glucuronidase (GUS)]. GUS signals from histochemical assays were found in various organs and tissues (Figure 3). The intensive staining in the hypocotyl and cotyledon veins of 7-d-old seedlings and in cotyledons and the margin serration tips of leaves of 14-d-old seedlings (Figure 3, A and B) is consistent with the high read coverage observed in the light-grown seedling in RNA-Seq data. A closer look at the shoot apex of 14-d-old seedlings further revealed elevated GUS activity in stipules. In young roots, staining was detected in stele tissues (Figure 3C). In rosette leaves, a spotted staining, corresponding with a strong GUS expression in trichomes and trichome basal cells was observed, as well as staining of the vascular tissues (Figure 3, D, E, and F). With ongoing development of the leaves, the GUS signal intensity gradually declined from the central region, causing the staining to be most intense close to the leaf margin and in tooth protrusions. The GUS signals observed in inflorescences point to increasing AtSNAPc4 promoter activity during pollen development (Figure 3, G and H). Weak GUS signals were further detected in young embryos (seeds) and silique abscission zones, as well as in cells of ruptured micropylar endosperm caps (Figure 3, I and J).

Figure 3.

Activity of the AtSNAPc4 promoter. Analysis of the expression patterns of uidA under the control of the AtSNAPc4 promoter by histochemical GUS staining. A, 7-d-old seedling, (B) 14-d-old seedling, (C) young root, (D) rosette leaf, (E) tip of a cauline leaf, (F) trichome, (G) inflorescence, (H) pollen, (I) young silique, and (J) empty seed coat.

The snapc4-1 and snapc4-2 mutants show defects in fertilization and embryogenesis

To be able to follow a reverse genetic approach, we aimed to identify AtSNAPc4 loss-of-function alleles. We identified seven independent T-DNA insertion mutant lines with insertions at the At3g18100 locus, designated snapc4-1–snapc4-7 (Figure 4A;Supplemental Table S1), all in Columbia-0 (Col-0) background. The zygosity of the T-DNA insertion allele was determined in plants of all insertion lines by polymerase chain reaction (PCR) genotyping. We were able to identify homozygous mutant plants for the lines carrying the alleles snapc4-3–snapc4-7. Homozygous plants of these lines did, however, not display apparent phenotypical abnormalities or transmission defects when grown under normal greenhouse conditions. We conclude that presumably due to the specific position, orientation or composition of the respective T-DNA insertion (Figure 4A), these alleles are either (very) weak or even close to wild-type (WT) activity. Real-time quantitative PCR (qPCR) analysis of AtSNAPc4 expression levels in seedlings of the respective homozygous mutant lines snapc4-3–snapc4-7 confirmed this conclusion (Supplemental Figure S4).

Figure 4.

Analysis of AtSNAPc4 T-DNA insertion mutants. A, Integration sites and T-DNA orientations for the At3g18100 T-DNA insertion mutant lines as determined by various PCR analyses. (B) Segregation of the snapc4-1 and the snapc4-2 alleles. TEs through the male and female gametophyte. χ²: result of the test of goodness-of-fit of the observed segregation ratio to the 1:2:1 ratio of Mendelian inheritance for selfed plants, respectively, 1:1:0 for the reciprocal crosses. C, Representative silique of a selfed snapc4-1/+ plant with aborted seeds. The siliques of Col-0 and snapc4-2/+ plants had no vacant positions (data not shown). Viability staining of pollen from Col-0 (D, E), snapc4-1/+ (F, G), and snapc4-2/+ plants (H, I). Viable pollen grains show intensive purple staining in the cytoplasm, aborted pollen grains are devoid of cytoplasm and only greenish staining of the exine outer layer is visible. Only ∼50% of the pollen of snapc4-1/+ plants was viable.

No homozygous mutants could be identified for the insertion alleles snapc4-1 and snapc4-2. Self-pollination of snapc4-1/+ and snapc4-2/+ hemizygous plants also yielded only hemizygous and WT offspring (Figure 4B). The genotypic analysis of the progeny of self-pollinated plants hemizygous in the snapc4-1 allele, resulted in a 50:77 (1:1.54) segregation of plants carrying the homozygous WT allele (+/+) and hemizygous plants. This non-Mendelian, distorted segregation ratio suggests that the snapc4-1 mutation most likely causes a defect in gametophytic functions. Self-pollination of the hemizygous snapc4-2 plants resulted in a 112:107 segregation of WT (+/+) and snapc4-2/+. This segregation ratio of 1:1.04 indicates a male or female gametophytic lethality. To determine how the mutations affect the male or female gametophytic functions, the snapc4-1/+ and snapc4-2/+ plants were used as male or female parent in crosses with WT plants. The transmission efficiency (TE) for the snapc4-1 allele was drastically reduced, with only ∼60% of either gametes successfully transmitting the mutant allele (Figure 4B). For the snapc4-2 allele, a reduced female TE of 80% was detected, while snapc4-2 pollen did not produce any progeny. These results suggest that the snapc4-2 mutation cannot be transmitted through the male gametophyte while the snapc4-1 mutation has a lower impact on male gametophytic function. The transmission through the female gametophyte is still possible, but reduced for both mutant alleles.

Seedset analyses in snapc4-1/+ plants revealed that ∼40% of the seeds were aborted and showed no sign of embryogenesis (Figure 4C), while seedset analyses in snapc4-2/+ plants did not show any apparent phenotypic abnormality. This indicates that the snapc4-2 allele does not have a strong impact on embryo development, but rather has a disruptive influence on an important step of the development preceding embryogenesis. Furthermore, results from pollen viability analyses with Alexander stain (Alexander, 1969) indicated about half of the pollen grains of snapc4-1/+ plants being unviable (Figure 4, F and G), whereas the pollen grains of snapc4-2/+ plants show no aberrant phenotype (Figure 4, H and I). The differing phenotypes of plants carrying snapc4-1 and snapc4-2 alleles might be caused by a residual function of the AtSNAPc4 protein produced by the snapc4-1 allele, while the snapc4-2 allele displays a more severe loss of function.

The A. thaliana genome encodes orthologs of all three core SNAPc subunits

Based on the result that our gene of interest encodes a distant ortholog of the HsSNAPc4 and DmSNAPc4 proteins, we analyzed the genome of A. thaliana for the occurrence of genes encoding the other SNAPc subunits, based on sequence homology of the translated proteins to the respective proteins of H. sapiens and D. melanogaster. We found the previously known AtSNAPc3/SRD2/At1g28560 gene, encoding a 386-aa protein with substantial sequence similarity to the two previously characterized SNAPc3 proteins in the C-terminal half (Supplemental Figure S5), where a SNAPc3-specific, unorthodox zinc finger domain (zf-SNAP50_C, pfam12252) is localized. In addition, we identified a SNAPc1 ortholog-encoding gene (At3g53270) which has not been characterized yet. The sequence similarity of the N-termini of HsSNAPc1, DmSNAPc1, and AtSNAPc1 is consistent with the location of the conserved SNAPc_SNAP43 domain (pfam09808) in all three proteins. Genes encoding orthologs of the subunits HsSNAPc2 and HsSNAPc5 were not identified in the A. thaliana reference genome sequence. This fits the assumption that these subunits play regulatory roles unique to mammalian or vertebrate systems (Mittal et al., 1999; Li et al., 2004).

The AtSNAPc components co-localize in the nucleus

Since orthologs of all core SNAPc subunits are encoded in the A. thaliana genome, we investigated whether the identified proteins are able to form an AtSNAPc4. In a first analysis, we used various localization prediction tools to determine the subcellular localization of the core SNAPc components. The sequences of AtSNAPc4 and AtSNAPc3 contain nuclear localization signals with high confidence scores, whereas the sequence of AtSNAPc1 does not. However, as AtSNAPc1 is predicted to be a small protein of only 33 kDa, localization into the nucleus by passive diffusion through nuclear pore complexes is feasible.

We experimentally examined the in vivo localization in transiently transfected tobacco BY-2 protoplasts. AtSNAPc4 was tagged N-terminally with cerulean fluorescent protein (CFP), while AtSNAPc1 and AtSNAPc3 were tagged C-terminally with enhanced yellow fluorescent protein (YFP) and monomeric red fluorescent protein (RFP) 1, respectively. Figure 5 shows the qualitative and quantitative analysis of the subsequent localization experiment. We calculated the Manders’ Overlap Coefficient (MOC) for all captured protoplast pictures and were able to distinguish between three types of co-localization (Figure 5A). While the protoplasts of all three types share a high co-localization of CFP-AtSNAPc4 signals with fluorescence signals of AtSNAPc1-YFP and AtSNAPc3-RFP, they differ in the MOC values calculated for AtSNAPc1-YFP and AtSNAPc3-RFP. Localization of CFP-AtSNAPc4 exclusively to the nucleus is consistent in all three types (Figure 5B). In detail, the detected signals indicate accumulation of the CFP-tagged AtSNAPc4 in the nucleolus and the nucleolar rim, as well as occasionally in spherical nuclear bodies and weakly in the nucleoplasm. In Type I protoplasts, AtSNAPc1-YFP is detected in the whole protoplast while AtSNAPc3-RFP is detected only in the nucleus, but the signal intensities of both fluorescent proteins are highest at the nucleolar rim. Type II protoplasts also feature YFP signals in the whole protoplasts and RFP signals solely in the nucleus, but the signal intensities indicate a higher degree of accumulation of the proteins in the same subnuclear regions as CFP-AtSNAPc4. About 50% of all captured protoplasts are categorized as Type III protoplasts, which feature almost perfect signal overlap in all three channels (Figure 5C). The observed localizations confirm the in silico localization predictions for AtSNAPc4 and AtSNAPc3, and verify the assumption that AtSNAPc1 can enter the nucleus by passive diffusion. The detected specific subnuclear co-localization patterns support the hypothesis that the three A. thaliana SNAPc subunit orthologs could interact at the molecular level and assemble into a protein complex.

Figure 5.

Co-localization analysis of the AtSNAP complex subunits. BY-2 tobacco protoplasts were co-transfected with CFP-AtSNAPc4, AtSNAPc1-YFP, and AtSNAPc3-RFP constructs, images were acquired with CLSM. A, The intensity-based MOC allowed the classification of the analyzed protoplasts into three categories. The error bars indicate the standard deviation of at least six independent measurements. B, Representative images of Types I–III co-localization protoplasts. Scale bars represent 5 µm. C, Fluorescence intensities correlated with pixel position, measured along the arrows in the according representative images. pp, protoplast; n, nucleus; no, nucleolus; nb, nuclear body.

AtSNAPc1 acts as molecular bridge between AtSNAPc4 and AtSNAPc3

To analyze whether the observed co-localization is caused by direct interaction of the components, we used the bimolecular fluorescence complementation (BiFC) technique, which enables the visualization of protein interaction in living cells. To this end, we generated constructs encoding the subunits AtSNAPc4, AtSNAPc1, and AtSNAPc3 tagged at the N-terminus with nonfluorescent fragments of YFP, either the N-terminal fragment (aa 1–155, YFP^N) or the C-terminal fragment (aa 156–239, YFP^C). Formation of a complex, guided by the association of complementary tagged proteins, restores the fluorescent properties of YFP. We included YFP^N and YFP^C fusions to the unrelated, nucleus-localized transcription factor AtMYB11 (Stracke et al., 2007) as negative controls in our experimental setup. This enabled the determination of noise levels originating from occasionally occurring unspecific and irreversible interactions between YFP^N and YFP^C (Horstman et al., 2014). Furthermore, a construct expressing RFP-tagged plasma membrane marker Syntaxin122 (Assaad et al., 2004) was included as a transfection control in every experiment. This allowed a distinction whether a protoplast exhibited no detectable YFP fluorescence because the two tested proteins did not interact, or because the examined protoplast was merely not transfected.

Figure 6 shows confocal laser scanning microscopy images of transfected tobacco BY-2 protoplasts with representative YFP intensities and localizations for all tested protein combinations. When testing the interaction of YFP^N/C-AtSNAPc4 and YFP^N/C-AtSNAPc1, we observed distinct fluorescence signals in the nucleus with an aggregation of the reconstituted YFP at the nucleolar rim and in the nucleolus. Co-transfection of YFP^N/C-AtSNAPc1 and YFP^N/C-AtSNAPc3 constructs also resulted in strong fluorescence signals, predominantly observed in the nucleoplasm. In protoplasts co-transfected with YFP^N/C-AtSNAPc4 and YFP^N/C-AtSNAPc3 constructs, we detected only very weak fluorescent signal intensities. Quantitative evaluation of the signal intensities showed the YFP signals in those protoplasts to be in the intensity range of protoplasts co-transfected with the negative control constructs (Supplemental Figure S6). The results of our BiFC experiment imply that AtSNAPc1 interacts with AtSNAPc4 and AtSNAPc3, while the subunits AtSNAPc4 and AtSNAPc3 seem not to interact directly. This indicates that the subunit AtSNAPc1 could serve as molecular bridge in the AtSNAPc.

Figure 6.

Pairwise BiFC analysis of the AtSNAPc components. Representative images from tobacco BY-2 protoplasts co-transfected with YFP^N and YFP^C fusion constructs, acquired with CLSM. In our experimental setup the expression was controlled by a CaMV 35S promoter and the YFP-fragments were N-terminal fused to the proteins of interest. The nuclear-localized transcription factor AtMYB11 served as negative control. Additionally, a RFP-SYN122 construct localizing to the plasma membrane was co-transfected as marker for successfully transfected protoplasts. Scale bars represent 5 μm.

AtSNAPc4 binds to snRNA gene promoters

The MYB domains of the HsSNAPc4 and DmSNAPc4 subunits have been shown to bind to the promoters of snRNA genes (Wang and Stumph, 1998; Mittal et al., 1999). Hence, we investigated whether AtSNAPc4 is able to bind to snRNA gene promoters as well. For this analysis, we chose the promoters of two of the Pol II-transcribed A. thaliana snRNA genes U1a (At5g49054) and U2.2 (At3g57645;Wang and Brendel, 2004). The USE and TATA box of A. thaliana snRNA promoters are located within 80-bp upstream of the transcription start site (Wang and Brendel, 2004). As the H. sapiens snRNA promoters contain a distal sequence element located at −200 to −250 (Baillat et al., 2012), we chose to work with 300-bp-long promoter fragments in order to include possible additional cis-regulatory elements.

We used the yeast one-hybrid (Y1H) system for DNA–protein interaction analysis. In our experimental setup the promoter DNA fragments were fused to the HIS3 reporter gene and incorporated into the genome of yeast strain Y187, generating DNA bait strains. A prey construct encoding a fusion of AtSNAPc4 to the GAL4 transcription activation domain (AD) was tested for binding to the snRNA gene promoters (Figure 7A). Positive interactions of the U1a and U2.2 snRNA gene promoters with the AD-AtSNAPc4 construct were observed. These results suggest that AtSNAPc4 is able to bind to A. thaliana snRNA gene promoters in vivo.

Figure 7.

Y1H assay analyzing the binding of AtSNAPc4 to snRNA promoters. For each tested Y1H interaction, yeast cells containing snRNA promoters as bait sequence were transformed with the prey plasmids (either the empty vector or AtSNAPc4). The transformed cells were plated on -His/-Trp selection medium supplemented with the competitive HIS3 inhibitor 3-AT of the indicated concentrations in three dilutions.

AtSNAPc4 co-localizes with the CB marker protein Atcoilin

In the co-localization and BiFC interaction assay analysis of the AtSNAPc subunits, we noted the aggregation of fluorescently tagged proteins in spherical nuclear bodies. As CBs were shown to be involved in snRNA gene transcription, we investigated whether the observed signals were caused by association of AtSNAPc components with CBs. The extensively studied CBs in nuclei of human cells are characterized by coilin (encoded by the COIL gene), the major structural scaffolding protein necessary for CB formation, structure, and activity (Machyna et al., 2015). In A. thaliana, a distant homolog was discovered and shown to be required for the formation of CBs in plant cells (Collier et al., 2006). Hence, we analyzed the co-localization of AtSNAPc4 and Atcoilin in BY-2 protoplasts by confocal microscopy. Figure 8 shows the localization of GFP-AtSNAPc4 and RFP-Atcoilin fusion proteins in three representative protoplasts. We observed both fluorescently tagged proteins to co-localize in the nucleolus, accompanied by either small numbers of large spherical nuclear bodies (Figure 8A) or large numbers of small nuclear bodies (Figure 8B). Moreover, we repeatedly detected a concentration of GFP-AtSNAPc4 in foci adjacent to the spherical bodies characterized by RFP-Atcoilin accumulation (Figure 8, C and D). The observed spatial relations of AtSNAPc4 and Atcoilin indicate a link between CBs and the SNAP complex in A. thaliana.

Figure 8.

Cellular localization of AtSNAPc4 and Atcoilin. Representative images from BY-2 protoplasts transfected with GFP-AtSNAPc4 and RFP-Atcoilin constructs, acquired with CLSM. All images share a co-localization of both proteins in the nucleolus but can be separated into three localization patterns in the nucleoplasm: (A) a small number of large CBs, (B) a large number of small CBs, and (C) GFP-AtSNAPc4 localizes in foci adjacent to CBs. Scale bars represent 5 μm. D, Fluorescence intensities correlated with pixel position, measured along the arrow in subfigure C.

Discussion

HsSNAPc4, the largest subunit of the human SNAP complex, was first identified in nuclear extracts of HeLa cells (Yoon et al., 1995) and subsequently an ortholog was described for D. melanogaster (Li et al., 2004). Both proteins were shown to be vital components of the snRNA gene transcription initiation machinery. The protein AtSNAPc3, encoded by the gene SRD2, has been identified as an ortholog of SNAPc3 and was characterized to play a crucial role in snRNA transcription activation (Ohtani and Sugiyama, 2005). Although it was mentioned that the A. thaliana genome contains genes encoding orthologs of the SNAPc4 and SNAPc1 subunits (Ohtani and Sugiyama, 2005; Ohtani, 2017), no further characterization of plant SNAP complex subunits was performed to date. We provide an overview of occurrence and structure of 4R-MYB/SNAPc4 genes and proteins in eukaryotic genomes and present insights into the function of AtSNAPc4—the A. thaliana SNAPc4 ortholog.

Alternative splice variants of AtSNAPc4 are possible NMD pathway targets

The A. thaliana SNAPc4 ortholog (synonym: AtMYB4R1) is encoded by the gene locus At3g18100, for which we identified five transcript variants (Figure 2), three of which are included in the Araport11 annotation (Cheng et al., 2017). An additional transcript variant (At3g18100.3/AtSNAPc4.3) was annotated in Araport11, but no experimental evidence for the existence of this variant has yet been established. In proteomic datasets of A. thaliana, we identified eight peptides that match the AtSNAPc4 protein sequence (Baerenfaller et al., 2008, 2011; Castellana et al., 2008; Durek et al., 2010). Six of these peptides match in regions exclusive to the translated protein sequence of splice variant 1. As all other established splice variants introduce reading frame shifts that result in PTCs, it is conceivable that these alternatively spliced transcripts are not translated into proteins at all, but rather trigger the nonsense-mediated mRNA decay (NMD) pathway that is known to degrade PTC-containing transcripts (McGlincy and Smith, 2008). Indeed, significantly elevated levels of AtSNAPc4 transcripts were detected in an NMD pathway mutant when compared to the control line (Gloggnitzer et al., 2014), indicating that a considerable amount of AtSNAPc4 transcripts are degraded via the NMD pathway.

AtSNAPc4 mutant alleles and expression

Analyses of T-DNA insertion mutants indicate that a lack of fully functional AtSNAPc4 influences gametophytes and zygotes, causing the absence of homozygous insertion mutants in the progeny of plants hemizygous for the strong alleles snapc4-1 and snapc4-2 (Figure 4). In line with these observations, no homozygous mutant mice carrying the Snapc4^{tm1a(KOMP)Wtsi} mutant allele were identified in a genome-wide gene knockout approach (Skarnes et al., 2011). In contrast, for zebrafish, a mutant line that is homozygous in the SNAPC4-G1018A null mutation allele is known (Voz et al., 2012). However, effects of this mutation do not occur before the third day after fertilization, due to maternal deposits of functional SNAPc4 transcripts or proteins that are sufficient for the development to proceed through embryogenesis (Voz et al., 2012). It is possible that homozygosity of the snapc4-1 and snapc4-2 alleles is also not cell lethal, but rather cause changes in the spliceosome that result in insufficient transmission through gametophytes. Indeed, loss-of-function mutations of a number of A. thaliana genes encoding spliceosomal components result in impaired transmission through male and female gametophytes (Park et al., 2019). A. thaliana mutants carrying the srd2-2 or srd2-3 T-DNA insertion alleles of the AtSNAPc3-encoding gene SRD2 (At1g28560) also failed to produce homozygous mutant plants, presumably due to defects in gametophyte and zygote development. In the progeny of self-pollinated snapc4-2/+ plants, we determined a 1:1 segregation ratio of WT and hemizygous plants, which is identical to the segregation ratio of self-pollinated srd2-2/+ plants (Ohtani et al., 2008). Additionally, we detected a similar percentage of aborted seeds in siliques born on snapc4-1/+ plants, as observed in seedset analyses of plants hemizygous in the srd2-2 allele (Ohtani et al., 2008). In contrast, reciprocal crosses of the hemizygous plants with WT plants produced different TEs for the mutant alleles. The srd2-2 allele was found to be transmitted mainly through the male gametophyte (TE_M 72%, TE_F 35%; Ohtani et al., 2008), whereas we noted a reduction to ∼60% TE of the snapc4-1 allele for female as well as for male gametophytes, and a complete failure to transmit the snapc4-2 allele via the male gametophyte.

The combination of shared and differing effects on the A. thaliana reproduction machinery is also reflected in somehow differential promoter activities that were observed in GUS reporter lines (Figure 3). While the promoters of AtSNAPc4 and SRD2 likewise are highly active in mature pollen, promoter activity in unfertilized pistils was only detected for the SRD2 gene (Ohtani et al., 2008). Analysis of the AtSNAPc4 promoter activity in seedlings further indicated high transcription rates in hypocotyl, root stele, and cotyledons, especially in vascular tissues. While the differing characteristics of the AtSNAPc4 and SRD2 transcriptional activities ad hoc seem to contradict the formation of a protein complex from the gene products, discrepancies in the transcriptional regulation of complex subunits are not unusual. Such a lack of (the level of) co-transcription can either be caused by additional functions of the subunits referred to as moonlighting (Jeffery, 1999) or be a mechanism that enhances correct complex assembly by counteracting affinity discrepancies of the subunits (Matalon et al., 2014). In fact, moonlighting is not unheard of for SNAPc core subunits. The HsSNAPc1 subunit was shown to have a direct role in the transcription of a large number of protein-coding genes (Baillat et al., 2012). Taken together, it is feasible that AtSNAPc3, AtSNAPc4, or both subunits might have additional roles beyond their contribution to AtSNAPc. However, a mutual role of the AtSNAPc component genes, AtSNAPc4, SDR2/AtSNAPc3, and AtSNAPc1, is supported by the relative low expression of all three genes in all analyzed RNA-Seq data from several databases. This weak expression in all tissues seems to be in good agreement with the SNAP complex core function in splicing.

The evolutionary conserved single-copy status of 4R-MYB genes supports the proposed functional conservation as SNAPc subunits

The MYB domain of AtSNAPc4 categorizes the protein as a special member of the MYB protein family. AtSNAPc4 is the only 4R-MYB protein encoded in the genome of A. thaliana (Stracke et al., 2001; Dubos et al., 2010). This singularity is an unusual feature as genes encoding other MYB-type proteins are usually heavily multiplied in plants (Paterson et al., 2006). We investigated whether the single-copy status of the 4R-MYB gene was specific for A. thaliana, and performed a survey for genes encoding 4R-MYB proteins in a broad range of 100 sequenced eukaryotic genomes, with a focus on land plants. In most of the analyzed genomes, we identified a single 4R-MYB gene (Figure 1), which indicates that the single-copy status is maintained throughout eukaryotic evolution. The conservation of the single-copy status in such evolutionary distant species is most likely due to selection against duplication, in which the genes are persistently returned to single-copy status following whole genome or small-scale duplication events (De Smet et al., 2013). Fittingly, functional analyses of conserved single-copy genes revealed that many of the encoded proteins are subunits of multiprotein complexes involved in essential functions (Koonin et al., 2004; De Smet et al., 2013; Han et al., 2014). It is assumed that those genes are repeatedly restored to a single copy after small-scale duplication events, because the gene products are dosage sensitive, in such a way that an imbalance of the corresponding protein complex subunits would have deleterious effects (Papp et al., 2003; De Smet et al., 2013; Han et al., 2014). Indeed, the phenotypes of mutants overexpressing AtSNAPc4 support the hypothesis that AtSNAPc4 is involved in dosage-sensitive processes as these mutants show a number of pleiotropic effects including epinastic leaves, loss of apical dominance, dwarfism, early flowering, twisted siliques, abscission defects, and severely altered flower and inflorescence morphologies (Supplemental Figure S7). Another explanation for the loss of duplicate gene copies after whole-genome duplication events can be found in the dominant-negative effect hypothesis (reviewed in De Smet et al., 2013). This hypothesis assumes that the restoration to single copy is a way to counteract the increased mutational target of duplicated genes, as mutations in one of the copies might produce dominant-negative phenotypes. The presence of single-copy genes encoding orthologs of the 4R-MYB SNAPc subunit in such a broad range of eukaryotic genomes, suggests a conservation of the snRNA transcription initiation machinery along the course of the eukaryotic evolution.

Surprisingly, the single-copy analysis also revealed the absence of a 4R-MYB gene in the well-defined genome of Saccharomyces cerevisiae. This does, however, not contradict the suggested functional conservation of 4R-MYB proteins, as a complete loss of a snRNA-specific transcriptional system in yeast was suggested previously (Ohtani, 2017), based on the absence of SNAPc subunit orthologs (Huang and Maraia, 2001) and the transfer RNA gene-like promoter structure of U6 snRNA genes (Hernandez, 2001).

Moreover, while most of the 60 analyzed plant genomes hold a single 4R-MYB gene, we identified two copies in the genomes of 12 embryophytes and three copies in the genome of A. lyrata. While the copy number discrepancy between the closely related Arabidopsis species seems peculiar, a phylogenetic study of Brassicaceae genome evolution showed major genome rearrangements in A. thaliana and large numbers of genes that were lost and gained since the species speciation event 12 million years ago, resulting in a different number of chromosomes (8 versus 5) and 4,275 more genes in A. lyrata than in A. thaliana (Murat et al., 2015). Six of the eight analyzed Brassicaceae family plant genomes encode multiple copies, indicating that shared genome duplication events like the alpha and beta tetraploidy (Bowers et al., 2003) might be the origins of these duplicates. Our data indicate that the third copy of A. lyrata located on chromosome 7 originates from a short segmental duplication (Supplemental Figure S2). Part of the coding sequence and the regions up- and downstream of the gene align to the At4g19630 locus and its genomic context, which matches the evolutionary relationship of A. thaliana chromosome 4 and A. lyrata chromosome 7 (Murat et al., 2015). Analyses of the phylogenetic relationships of the other 4R-MYB gene copies and of the occurrence of duplicates in relation to known embybryophyte polyploidization events (Supplemental Figure S1) evince the majority of the 4R-MYB gene pairs stem from species-specific duplication events. It is feasible that the duplicates in these genomes have not yet been restored to single-copy status.

4R-MYB proteins share additional characteristics besides the MYB domain

Within the analyzed set of SNAPc4 orthologs, the different proteins vary considerably in protein length and MYB domain position within the protein (Figure 1). Furthermore, our MSA analysis showed high sequence diversity for the whole-length proteins combined with a strong conservation of the MYB domain (Supplemental Figure S3B). We additionally identified a 28 aa long SNAPc4 heptad motif (Supplemental Figure S3, A and C) in most of the orthologs. Such a seven-residue periodicity (abcdefg_n) is a characteristic sequence feature of coiled-coil protein domains. Coiled coils comprised two or more α-helices wrapped around each other forming compact helical bundles and are known to function as protein–protein interaction site in protein complex assemblies (Strauss and Keller, 2008). The a and d positions of the coiled-coil heptad motif are predominantly occupied by hydrophobic residues (Lupas, 1996), although a recent study observed that the a and d positions of ∼25% of the coiled-coil structures deposited in the CC+ database of coiled coils (Testa et al., 2009), are occupied by polar residues (Thomas et al., 2017). A closer examination of the conserved residues of the SNAPc4 heptad motif shows the a and d positions to be either occupied by hydrophobic residues or the polar aa asparagine and glutamine (Supplemental Figure S3C), which further underpins the suspicion of the SNAPc4 heptad motif being a coiled-coil domain. Moreover, the SNAPc4 heptad motif in HsNSAPc4 is localized in a region known to directly interact with HsSNAPc1 (Ma and Hernandez, 2001). These findings suggest a general involvement of the conserved SNAPc4 heptad region in the interaction of SNAPc4 and SNAPc1 subunits, potentially realized by coiled-coil structures. In contrast, the position of the SNAPc4 heptad motif in DmSNAPc4 (aa 33–60) contradicts this guess, as a truncated DmSNAPc4 protein from aa 176–451 was determined to be sufficient for binding to DmSNAPc1 and a truncation of the first 62 aa of the DmSNAPc4 N-terminus did not diminish the proteins ability to bind DmSNAPc1 (Hung et al., 2009). However, there are distinct differences in the subunit interaction of the two previously characterized SNAP complexes. In the case of DmSNAPc, all three subunits were shown to interact with each other (Hung et al., 2009), whereas the HsSNAPc1 subunit was demonstrated to function as molecular bridge between HsSNAPc4 and HsSNAPc3 (Mittal et al., 1999), which is a functional property we also observed for the AtSNAPc1 subunit in our study. More investigations are necessary to determine if an evolutionarily conserved association pattern exists and whether the identified SNAPc4 heptad motif is involved in protein–protein interaction.

Further analysis of the protein sequences of the orthologs revealed the aggregation of regions rich in aspartic and glutamic acid at the N-termini of the proteins (Figure 1), especially in land plants and algae. D/E-rich regions, in general, are known to function primarily as DNA mimics, in mRNA processing, and in transcription complex regulation, though specific D/E-rich regions were observed to function in metal binding (Chou and Wang, 2015). Of particular interest in this context are the D/E-rich regions of the yeast proteins TAF1 and Brf1, which have been shown to competitively bind to the same surface groove of the yeast TATA-box binding protein (TBP; Juo et al., 2003; Anandapadamanaban et al., 2013). For the HsSNAPc4 protein, a 50 aa long D/E-rich TBP recruitment region 1 (aa 34–84) was identified, which is essential for the binding of the mini-SNAPc to TBP and located in the D/E-enriched N-terminus (Ma and Hernandez, 2002). Since SNAP complexes as well as TBPs are essential parts of the transcription initiation complex accumulating at snRNA gene promoters, it is plausible to assume that the identified D/E-rich regions, and the D/E-rich N-termini in general, are of importance in the interaction of SNAP complexes with TBPs.

Subnuclear localization

Our localization analysis of the fluorescently tagged A. thaliana SNAP subunits revealed that the proteins co-localize in the nuclei of tobacco BY-2 protoplasts (Figure 5), which is consistent with the hypothesized function in snRNA transcription. In detail, we observed the tagged proteins often co-localize at the rim of nucleoli and spherical nuclear bodies. The aggregation at the nucleolar rim might be caused by localization of the AtSNAPc subunits to the granular component (GC). The nucleolus consists of three morphologically distinguishable, evolutionary conserved regions: fibrillar centers, the dense fibrillar component, and the enveloping GC that is composed of densely packed granules of RNA and proteins (Shaw and Jordan, 1995). Formation of a thin ring of GC surrounding the nucleolus is characteristic for cells in cell cycle phase G1 (González-Camacho and Medina, 2006). Historically, the nucleolus is seen as ribosome factory, in which the final steps of ribosome biogenesis take place in the GC (Hadjiolov, 1984). However, the precise functions of the nucleolar domains are still debated and functions of the nucleolus in many other key cellular activities have been shown since (reviewed in Kalinina et al., 2018).

Besides the co-localization at the nucleolar rim, we also repeatedly observed the tagged AtSNAPc subunits in nucleolar spherical bodies of varying size and number. Based on published studies that functionally and locally link snRNA transcription to CBs (Dundr et al., 2007; Wang et al., 2016), we analyzed the localization of the AtSNAPc4 subunit relative to the localization of the CB marker protein Atcoilin. The results obtained from this experiment clearly show a co-localization of the two proteins in many small or few large nuclear bodies (Figure 8), which matches the inverse correlation of number and size known for plant CBs (Acevedo et al., 2002). All examined protoplasts also featured a prominent localization of Atcoilin to the nucleolus, which has been reported before for other coilin orthologues as a consequence of overexpression and/or experimental conditions (Hebert and Matera, 2000; Trinkle-Mulcahy and Sleeman, 2017). Another striking observation of the co-localization analysis was the frequent aggregation of AtSNAPc4 proteins in foci adjacent to the CBs. In H. sapiens HeLa cells, CBs have been shown to promote aggregation of snRNA loci in their proximity (Wang et al., 2016), thus the accumulation of AtSNAPc4 adjacent to CBs further supports the proposed functional role of AtSNAPc in snRNA transcription.

The proposed AtSNAPc subunits form a multiprotein complex and are linked to snRNA gene transcription

Using BiFC, we confirmed that the observed co-localization of the AtSNAPc subunits in the nucleus does originate from protein–protein interactions, and thus further underpin that AtSNAPc4, AtSNAPc1, and AtSNAPC3 form a protein complex (Figure 6). In detail, AtSNAPc1 functions as a molecular bridge between the subunits AtSNAPc4 and AtSNAPc3, analogous to the subunit assemblage of the human Mini-SNAPc (Mittal et al., 1999). The fluorescence signals of re-assembled YFP molecules were found invariably in the nucleus, consistent with the previously observed nuclear co-localization of the tagged proteins.

We were able to show that addition of an AtSNAPc4 effector construct resulted in elevated expression levels of snRNA promoter-driven reporter genes in yeast cells, thus demonstrating the proposed binding of AtSNAPc4 to the tested A. thaliana Pol II promoters (Figure 7). In accordance with these results, AtSNAPc3/SRD2 was shown to function in snRNA transcription activation as well (Ohtani and Sugiyama, 2005; Ohtani et al., 2008). Based on these observations, it is a reasonable conclusion that the evolutionary conserved functional role of HsSNAPc and DmSNAPc in snRNA transcription initiation is shared by the AtSNAPc4.

In summary, our results show that the A. thaliana SNAPc subunit orthologs assemble into a SNAP complex that is involved in snRNA transcription and that 4R-MYB/SNAPc4 proteins are conserved along the course of eukaryotic evolution. It will be interesting to see whether other functions of the SNAP complex, like the role in chromatin architecture at snRNA gene loci in humans (O’Reilly et al., 2014), are conserved in plants as well.

Materials and methods

Plant materials

All A. thaliana lines used in this study were in Col-0 background. The At3g18100 T-DNA insertion lines (Supplemental Table S1) were identified from the GABI-Kat (Kleinboelting et al., 2012) and SALK (Alonso et al., 2003) collections and ordered from the Nottingham Arabidopsis Stock Center. Allele zygosity and T-DNA border orientations were determined by PCR-based genotyping assays as described by Bolle et al. (2013) using the primers listed in Supplemental Table S3.

Growth conditions

For germination, seeds were surface-sterilized and sown on half-strength Murashige and Skoog (MS) medium agar (0.5 × MS, 1% sucrose, and 0.8% agar; Murashige and Skoog, 1962). After stratification for 2–3 d in the dark at 4°C, the planted seeds were transferred to a phytochamber with long-day photoperiod conditions (16 h light/8 h dark) at 22°C. About 14 d after germination, the seedlings were transplanted to compost [60% white peat 0–8 mm, 30% vermiculite 2–3 mm, 10% sand, 1.4 g/L iron chelate, 2.1 g/L Radigen slow-release micronutrient fertilizer (2% iron, 0.8% molybdenum, 1.5% copper, 0.6% boron, 1% manganese, and 0.5% zinc; Terraflor), pH 5.5–6.0; Stender] and grown in a greenhouse under long-day conditions with a mean temperature of 21°C. Seeds were harvested ∼8 weeks after germination.

CDS and promoter cloning

Genomic DNA was isolated using the MasterPure Plant Leaf DNA Purification Kit (epicenter). Total RNA was isolated using the peq-GOLD TriFast kit (Peqlab Biotechnologie, Erlangen, Germany) and after gDNA removal using DNAse I (New England Biolabs, NEB), cDNA was synthesized using ProtoScript First-Strand cDNA Synthesis Kit (NEB). PCRs were performed with Q5 polymerase (NEB) according to manufacturer instructions, using A. thaliana Col-0 genomic DNA or pooled cDNA as template. The primers used to amplify the CDSs of AtSNAPc4, AtSNAPc1, AtSNAPc3, and Atcoilin and the promoter fragments of AtSNAPc4, U1a snRNA, and U2.2 snRNA genes, are listed in Supplemental Table S3. All constructs were produced using Gateway cloning technology, according to manufacturer instructions (ThermoFisher Scientific, Waltham, MA, USA) and verified by Sanger sequencing. The Gateway vectors used and the generated constructs are listed in Supplemental Table S2.

Stable plant transformation

Arabidopsis thaliana WT plants were transformed by the floral dip method (Clough and Bent, 1998) using Agrobacterium tumefaciens strain GV3101 carrying the helper plasmid pMP90 (Koncz and Schell, 1986) transfected with the proAtSNAPc4:GUS or the pro35S:AtSNAPc4 construct. Transgenic plants were selected on 0.5× MS agar plates containing 20 mg/L hygromycin B or by spraying a selection solution containing 240 mg/L glufosinate, respectively. AtSNAPc4-overexpression mutants carrying the pro35S:AtSNAPc4 construct were validated by PCR, the primers are listed in Supplemental Table S3.

Histochemical analysis of GUS reporter lines

Plant materials were fixed in 0.3% v/v formaldehyde, 10 mM MES/KOH pH 5.6, 300 mM mannitol, and subsequently rinsed with 50-mM sodium phosphate buffer (pH 7.0). The fixed samples were vacuum infiltrated with X-Gluc solution (0.05% (w/v) 5-bromo-4-chloro-3-indolyl-β-d-glucuronide in 50-mM sodium phosphate buffer) for 30 min, and incubated overnight at 37°C. Chlorophyll bleaching was accomplished by repeatedly washing the samples in ethanol. Three transgenic lines with resembling GUS staining patterns were chosen for further analysis.

Pollen viability staining

For the differential staining of aborted and nonaborted pollen, flower buds were dissected under a microscope and anthers were placed on microscopic slides in Alexander staining solution (Alexander, 1969). The slides were briefly heated on a hotplate before photographic documentation.

Localization of fluorescent proteins

Tobacco BY-2 cells were harvested from suspension culture medium 3 d after inoculation and protoplasts were isolated as described by Merkle et al. (1996). Protoplast recovery and the transient transfection using PEG were performed according to the specifications of Negrutiu et al. (1987). For the multicolor localization of fluorescent proteins, 40 µg of each fusion protein-encoding plasmid were co-transfected. For BiFC analysis, 20 µg of pro35S:YFP^N-X and pro35S:YFP^C-X constructs (see Supplemental Table S2) were co-transfected. Additionally, 3 µg of a plasmid encoding the plasma membrane marker Syntaxin122 fused to RFP (Assaad et al., 2004) were co-transfected in the BiFC experiments to serve as positive marker for successfully transfected protoplasts. After 20-h incubation in the dark, 50 µL of each transfection reaction were transferred into the well of an uncoated hydrophobic 8-well slide (Ibidi) and mixed with 500 µL W5 solution (154 mM NaCl, 125 mM CaCl_{2 *} 2H₂0, 5 mM KCl, 5 mM glucose, and 5 mM MES pH 5.7).

Transfected protoplasts were observed using a 63-fold magnification objective (Zeiss LCI Plan-NEOFLUAR, water-immersion, NA = 1.3) on a Carl Zeiss LSM780 inverted confocal laser scanning microscope. For the multicolor localization analysis of CFP, eYFP, and mRFP1 fusion proteins, the main bean splitter MBS 458/514/594 was used, allowing fast sequential excitation and thus line by line recording. CFP was excited with the 458 line of an argon–ion laser and detected in the range of 460–500 nm; eYFP was excited with the 514 nm line of the argon–ion laser and detected after passage through a 515–570 nm bandpass filter; mRFP1 was excited using the 594 nm line of a helium–neon laser and detected in the range of 600–700 nm. For BiFC analysis, the same excitation/bandpass settings were used for the eYFP and mRFP1 channel. Additionally, the detector gain of the eYFP channel was kept fixed for all analyzed samples, to allow comparative quantification of the fluorescence intensities. In the multicolor localization analysis of mGFP6 and mRFP1 fusion proteins, MBS 488/561 was used. mGFP6 was excited with the 488 nm line of the argon–ion laser and detected in the range of 500–550 nm, mRFP1 was excited with a diode-pumped solid state DPSS 561 laser and detected after passage through a 570–625 nm bandpass filter.

Images were acquired with the Zen software (Zeiss). Editing and line plot analyses were performed with Zen and the biological-image analysis software Fiji (Schindelin et al., 2012). When the localization of the fluorescence signal (rather than the signal intensity) was of interest, the highest observed signal for each channel was set as the maximum of the color gradient, thus correcting differences due to fluorophore traits. For quantitative evaluation, intensity thresholds for background signal exclusion were determined for each observed protoplast by application of Otsus’ threshold clustering algorithm (Otsu, 1979) in Fiji (Schindelin et al., 2012). The quantitative evaluation of co-localization is based on MOC and was calculated using the ImageJ plugin BlobProb (Fletcher et al., 2010). The average signal intensities of the protoplasts captured in the BiFC experiment was calculated with a custom ImageJ macro based on the remaining pixels after background subtraction.

Y1H assay

For Y1H analyses the yeast strains AH109 (James et al., 1996) and Y187 (Fromont-Racine et al., 1997) of opposite mating type were utilized. Both strains were transformed using LiAc, PEG, and single-stranded carrier DNA as described by Gietz and Schiestl, (2007). The proADH1:AD-AtSNAPc4 construct (in the Gateway compatible vector pDEST22 with GAL4 AD) and the empty vector control were transformed into the AH109 strain. Using the vector pMW#2, the promoters of the U1a and U2.2 snRNA genes were fused to the HIS3 reporter gene and integrated in the Y187 genome, generating “DNA bait” strains. For each DNA bait, multiple colonies were picked and spotted on SD-His plates with 0–50 mM 3-amino-1,2,4-triazole (3-AT) to test the strength of the HIS3 reporter gene self-activation. Colonies that exhibited minimal HIS3 self-activation were chosen and used in the subsequent interaction experiments. The proADH1:AD-AtSNAPc4 construct and the empty vector control were introduced into the chosen DNA bait strains by mating. The resulting diploid strains were grown in liquid SD-Trp/-His medium into stationary phase. The cultures were adjusted to an OD₆₀₀ of 0.5 and spotted as 10-fold serial dilution on SD-Trp/-His plates containing 0–60 mM 3-AT. Plates were incubated at 30°C for 5 d before pictures were taken. Yeast that showed higher reporter gene activity, that is, exhibited growth on high 3-AT concentrations, was regarded as positive indicators for protein–DNA interactions.

Supplemental data

Supplemental data are available in the online version of this article.

Supplemental Figure S1: Polyploidization events and number of 4R-MYB genes in selected land plant species.

Supplemental Figure S2: Schematic representation of local alignments of the three 4R-MYB genes from A. lyrata to the genome of A. thaliana.

Supplemental Figure S3: Sequence alignments of SNAPc4 orthologs.

Supplemental Figure S4: AtSNAPc4 expression analysis in seedlings of the (phenotypically) weak homozygous T-DNA insertion mutant lines snapc4-3 to snapc4-7.

Supplemental Figure S5: Comparison of ortholog SNAPc subunits of A. thaliana, H. sapiens and D. melanogaster.

Supplemental Figure S6: Quantitative evaluation of the signal intensities measured in the BiFC experiment, testing interactions of AtSNAPc subunits.

SupplementalFigure S7: Pleiotropic phenotype of pro2x35S:AtSNAPc4 overexpression lines.

Supplemental Table S1: A. thaliana lines with a T-DNA insertion at the At3g18100 locus.

Supplemental Table S2: Generated constructs.

SupplementalTable S3: Oligonucleotides used in this study.

Supplemental Table S4: List of identified 4R-MYB genes.

Supplementary methods

Relative expression analysis.

 

R.S. and B.W. conceived and supervised the project; K.T. and R.S. designed the experiments; K.T. performed the experiments and analyzed the data; K.T. wrote the manuscript draft; R.S. and B.W. supervised and completed the writing; R.S. agrees to serve as author responsible for contact and ensures communication.

The authors responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instruction for Authors (https://academic.oup.com/plphys) is: Ralf Stracke (ralf.stracke@uni-bielefeld.de).