Expression of HA-EgWRI1 rescues the low oil content in wri1-1 seeds

A comparison of protein sequences between AtWRI1 and EgWRI1 indicates that EgWRI1 is 93 amino acids shorter compared to AtWRI1 (Figure 1A). Almost all regions of high similarity between AtWRI1 and EgWRI1 are located in the N-terminal ~ 230 amino acids (Figure 1A). The C-terminal regions of AtWRI1 (190 amino acids) and EgWRI1 (98 amino acids) are strikingly diverged, (only 28% identical) with several large “deletions” within the EgWRI1 sequence. Furthermore, the predicted secondary structures bear little resemblance; AtWRI1 includes six helix regions in the C-terminal end of the protein, whereas EgWRI1 possesses no predicted helix regions in the C-terminal region (Figure 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Alignment of protein sequence of EgWRI1 and AtWRI1.

A) EgWRI1 is 93 amino acids shorter compared to AtWRI1. Most of the difference between protein sequences of EgWRI1 and AtWRI1 occurs at the C-terminal half of the protein (highlighted by boxes). Conservation of amino acids is denoted by different colors as illustrated by the scale bar. The alignment was analyzed by the PRALINE program (http://www.ibi.vu.nl/programs/pralinewww/). B) Secondary structure of AtWRI1 and EgWRI1 was analyzed by SWISS-MODEL (http://swissmodel.expasy.org/) and structure figure is manually generated based on the prediction. The orange cylinders and purple arrows indicate helix and extended-beta strands, respectively. Numerous helix and extended-beta structures that are found in the C-terminal of AtWRI1 are missing in the C-terminal of EgWRI1.

https://doi.org/10.1371/journal.pone.0068887.g001

To better understand features of the WRI1 structure that are important for WRI1 function we tested the ability of the divergent EgWRI1 to restore seed oil and other phenotypes in the Arabidopsis wri1-1 mutant background. Beyond EgWRI1’s ability to restore known wri1-1 phenotypes, we also asked whether its expression might lead to additional phenotypes, such as higher oil accumulation. Such a result might occur if a heterologous protein ‘escapes’ repression that possibly acts on the C-terminus of the native protein or its mRNA.

In order to investigate the ability of EgWRI1 to restore several phenotypes observed in wri1-1, we generated transgenic plants expressing HA-EgWRI1 under the constitutive cauliflower mosaic virus (CaMV) 35S promoter. As a positive control, experiments were conducted in parallel with transgenic wri1-1 expressing HA-AtWRI1. As shown in Figure 2, homozygous transgenic wri1-1 expressing HA-EgWRI1 displayed a ‘non-wrinkled’ seed surface and normal seed shape indicating rescue of these phenotypes (also observed in transgenic wri1-1 expressing HA-AtWRI1). We further measured the fatty acid content in seeds of wri1-1 expressing HA-EgWRI1. As shown in Figure 3, expression of HA-EgWRI1 restored the fatty acid content of wri1-1 seeds, similar to wri1-1 seeds rescued by expression of HA-AtWRI1. The fatty acid content in seeds of HA-EgWRI1 transgenic lines (#1-4, #11-1, #13-5 and #14-1) compared to WT were not different with statistical significance (P > 0.05, t-test). In addition to its lower oil content, the fatty acid composition of wri1-1 differs from WT in its lower relative content of 18:1 and higher content of 22:1 [1]. When the wri1-1 mutant is complemented with HA-AtWRI1 the fatty acid composition returns to a WT profile (Figure S1). In contrast, the fatty acid profile of wri1-1 expressing HA-EgWRI1 was only partially restored. Although, most fatty acids were not significantly different, the content of 18:1 and 22:1 were more similar to wri1-1 than to WT (Figure S1). A similar result was observed in the complementation of wri1 by ZmWRI1 orthologs [11].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. HA-EgWRI1 (or HA-AtWRI1) complements the “wrinkled” feature of wri1-1 seeds.

https://doi.org/10.1371/journal.pone.0068887.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. HA-EgWRI1 rescues the reduced oil phenotype of wri1-1 seeds.

Total fatty acids in seeds of WT, wri1-1 and transgenic wri1-1 expressing HA-EgWRI1 or HA-AtWRI1 are shown in the figure. Results are shown as means ± SE (n = 3-4) of biological replications.

https://doi.org/10.1371/journal.pone.0068887.g003

Seed germination and impaired seedling phenotypes of wri1-1 are restored by HA-EgWRI1

The wri1-1 mutant exhibits reduced seed germination and impaired seedling establishment when germinated on medium without the addition of sucrose [13]. Expression of HA-EgWRI1 corrected the reduced seed germination of wri1-1 (Figure 4A). Transgenic wri1-1 mutants expressing HA-EgWRI1 established seedlings normally on agar-solidified medium lacking sucrose (Figure 4B) and on soil (Figure 4C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Seed germination and seedling establishment of wri1-1 transformed with HA-AtWRI1 or HA-EgWRI1.

A) Expression of HA-EgWRI1 rescues the reduced germination of wri1-1 seeds. Results are shown as means ± SE (n = 3). Seed germination of all HA-EgWRI1 transgenic lines were not significantly different compared to WT (P > 0.05, t-test). B) Expression of HA-AtWRI1 or HA-EgWRI1 in wri1-1 complements the failure of seedling establishment in growth medium without sucrose. C) HA-EgWRI1 rescues impaired seedling establishment of wri1-1 plants in soil. Results are shown as means ± SE (n = 4). Seedling establishment of HA-EgWRI1 transgenic lines (#1-4, #11-1, #16-3) did not differ significantly compared to WT (P > 0.05, t-test).

https://doi.org/10.1371/journal.pone.0068887.g004

As a further test of the influence of WRI1 structural variants we also generated transgenic wri1-1 expressing EgWRI1-TAP or AtWRI1-TAP. The low fatty acid content (Figure S2) and altered fatty acid profile of wri1-1 seeds (Figure S3), reduced seed germination (Figure S5A) and impaired seedling establishment (Figure S4B) were all restored by AtWRI1-TAP. These results indicate that the fusion of the TAP tag at the C-terminus of AtWRI1 did not cause any interference with the AtWRI1 functions. In contrast, although the reduced seed germination of wri1-1 was restored by EgWRI1-TAP (Figure S4A), this construct only partially restored the fatty acid content (Figure S2) and profile (Figure S3) of wri1-1. In addition, seedling establishment remained unsuccessful in transgenic wri1-1 plants expressing EgWRI1-TAP (Figure S4B).

Small exon “VYL” is essential for function of AtWRI1

The results above indicated that EgWRI1 is able to complement the phenotypes of wri1-1 mutant despite major differences in the C-terminal amino acid sequences of WRI1 from Arabidopsis and oil palm. A detailed analysis of features conserved between AtWRI1, EgWRI1 and other WRI1 orthologs from diverse species revealed a short protein sequence “VYL” (Figure S5A) present in the first AP2 domain (responsible for DNA-binding [23,24]). To better understand the function of the VYL sequence we asked if amino acid changes of "VYL" would lead to an impairment of AtWRI1 function. A total of four versions of AtWRI1 with amino acids substitutions (AtWRI1^{V99A/Y100A/L101A}; AtWRI1^V99D; AtWRI1^Y100C; AtWRI1^L101Q) were generated and used to transform the wri1-1 mutant. Measurement of fatty acid content in these transgenic wri1-1 seeds indicated that AtWRI1^{V99A/Y100A/L101A} failed to restore the fatty acid content of wri1-1. The mutants AtWRI1^Y100C, or AtWRI1^L101Q, with single changes in the VYL sequence could partially, but not completely complement wri1-1 fatty acid content (Figure 5). Taken together, these results indicate that mutation of residues encoded by this small exon lead to impairment of AtWRI1 function and that “VYL” is an essential component of the AtWRI1 structure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Fatty acid content of seeds of WT, wri1-1 and transgenic wri1-1 expressing HA-AtWRI1 or HA-AtWRI1s mutated in VYL sequence.

Results are means ± SE (n = 3). “*” indicates significant difference (P<0.05, t-test) between transgenic wri1-1 lines expressing HA-AtWRI1 (lines #1 and #6), and lines expressing the mutated genes.

https://doi.org/10.1371/journal.pone.0068887.g005

Analysis of WRI1 alternative splice forms

In AtWRI1, “VYL” is encoded by a 9 bp exon (Figure 6). Micro-exons of 2–25 bp are known to sometimes facilitate alternative splicing events [25]. AtWRI1 is predicted to have three alternative splice forms (Figure 6) of different lengths and different protein sequences. Notably, “VYL” is absent from one of the predicted splice forms (At3g54320.2) In addition, Multiple Sequence Alignment examination of predicted protein sequences from approximately 34 plant genomes in the Phytozome database 9.0 (http://www.phytozome.org) indicated that VYL is also missing from the predicted AtWRI1-orthologous amino acid sequences of 13 species (Figure S5B). These results indicate that the prediction of alternative protein sequences (or splice forms) for WRI1-like proteins occurs throughout the plant kingdom, and is not peculiar to Arabidopsis or its close relatives. We also note that 2 of 34 predicted amino acid sequences of WRI1 orthologs at Phytozome include “IYL” in place of “VYL”.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Three predicted alternative splice forms of AtWRI1 (At3g54320.1, At3g54320.2, and At3g54320.3).

A) Model of AtWRI1 splice forms 1, 2 and 3. The notable features of these three forms are highlighted: (1) presence in splice forms 1 and 3 of a short 9 bp exon (TTTATCTGG) that encodes amino acids “VYL”; (2) presence of an intron at the 3’ end in splice form 1 and its absence in splice forms 2 and 3 is circled. B) Alignment of predicted protein sequence of three AtWRI1 alternative splice forms. Exon 3 (amino acids sequence “VYL”) in AtWRI1 splice forms 1 and 3 and absence from splice form 2 are highlighted in green. Unique amino acids “FQGLFVGSE” (form 1) and “V” (form 2 and 3) at the C-terminus of AtWRI1 splice forms are highlighted in yellow and red, respectively.

https://doi.org/10.1371/journal.pone.0068887.g006

The three predicted alternative splice forms of AtWRI1 (At3g54320.1, At3g54320.2 and At3g54320.3; http://www.arabidopsis.org) are referred to as splice form 1, 2 and 3 in this work. As highlighted in Figure 6, the distinguishing features of these three forms are: A) there is a short 9 bp exon near the N-terminus that encodes the amino acid sequence VYL. This exon is present in forms 1 and 3, but absent in form 2. B) In form 1, but not 2 or 3 there is an additional intron at the 3’ end of the coding sequence (circled). If removed by splicing, the stop codon is altered resulting in an additional 9 amino acids (FQGLFVGSE) at the C-terminus of form 1. C) Finally, in form 2, the first exon begins with a downstream ATG resulting in a protein that is 74 amino acids shorter. In the case of AtWRI1, the three splice forms are based on predicted gene models, rather than experimental evidence, but each has been previously assigned the same confidence level.

These predictions for Arabidopsis and for at least 33 other plant species with sequenced genomes, raised the question whether different WRI1 splice forms are present in vivo and have different functions. We specifically asked: 1) Are all three AtWRI1 splice forms expressed in Arabidopsis? 2) Can we find multiple WRI1 splice forms in other plant species? 3) If we can find multiple WRI1 splice forms, are these expressed differently in different tissues? To evaluate the possible expression of multiple spice forms of WRI1, we tested directly by reverse transcription polymerase chain reaction (RT-PCR), performed RNASeq analysis of developing Arabidopsis seeds, and also searched publicly available large transcript databases.

AtWRI1 splice form 3 is the major form expressed in Arabidopsis seedlings

In many cases the presence of alternative splice forms can be detected by the different sizes of RT-PCR products [18,22]. We therefore performed RT-PCR on RNA extracted from young Arabidopsis seedlings (3-to-9-day-old) and with primers targeted at the short 9 bp exon and the last 3’ intron. Because expression of AtWRI1 is found to be sugar-inducible [3], we grew Arabidopsis on medium with or without the addition of sucrose, to determine if different AtWRI1 splice forms might be expressed differently under these two conditions. A primer set (FW1 + RV1) was designed to amplify a PCR product of 157bp (in splice form 1 or 3) or 148 bp (in splice form 2) respectively (Figure S6A). Sequencing of fourteen independent PCR products (Figure S6B) indicated in all cases, the presence of AtWRI1 exon 3.

The primer set (FW2 + RV2) was designed to amplify a PCR product of 269 bp (in form 2 or 3) or 167 bp (in form 1) respectively. As shown in Figure S6C, all the samples that we tested supported the existence of form 2 or 3 (a PCR product of 269bp). A PCR product of form 1 (167bp) was not detected. Further sequencing of each PCR product in Figure S6C also confirmed this conclusion. Combining the results, this evidence indicated that splice form 3 of AtWRI1 is the major form expressed in Arabidopsis seedlings.

As a second approach to evaluate expression of alternative splice forms in Arabidopsis, we searched a number of databases of Sanger ESTs, 454 pyrosequencing and Illumina RNASeq. Although large EST and RNASeq databases are available from several Arabidopsis tissues, the datasets do not include large numbers of reads from developing seeds, where WRI1 is most highly expressed. Therefore, we performed RNASeq analysis on three stages of developing seeds. Approximately 100 million 50 nt Illumina reads were mapped to the TAIR10 genome and analyzed. From these data, WRI1 was expressed at a level of 165 Fragments per Kilobase per Million fragments (FPKM) (average of three developmental stages). This represents approximately 0.01% of the mRNA population and provides a more quantitative estimate of WRI1 expression than previously available for Arabidopsis developing seeds. Over 500 WRI1 reads mapped to the region of the WRI1 gene that overlaps the 9 bp exon 3. A section of the reads is presented in GBrowse format (Figure S7). No reads were detected that lacked exon 3. We also examined the 3’ sequences and found no sequences for splice form 1. Based on the number of reads representing splice form 3 and no reads representing forms 1 or 2, a binomial test indicated high probability (P > 0.999) that splice forms 1 or 2 of WRI1 are not expressed in Arabidopsis developing seeds.

Alternative splice forms of genes are sometimes expressed only in specific tissues [20,21]. To evaluate this possibility, we also examined publicly available Ilumina RNASeq data from Arabidopsis roots and flowers. Although WRI1 is expressed at much lower levels in these tissues (13.1 FPKM root, 10.9 FPKM flower), the 9 bp exon 3 was represented by over 80 RNASeq reads from roots and 20 reads from flowers. No RNASeq reads were found that lacked exon 3. Examination of the 3’ end (represented by 130 RNASeq reads in roots and 60 reads in flowers) revealed that the predicted 3’ intron was not spliced out in any reads. Thus, splice form 1 was also not detected in either root or flower datasets. Taken together, although expression of the alternative splice forms at very low levels cannot be ruled out, these data indicate that splice forms 1 or 2 are unlikely to play a biological role in Arabidopsis flowers and roots grown under standard condition.

Analysis of WRI1 alternative splice form expression in plant species other than Arabidopsis

The presence/absence of the 9 bp short exon in WRI1 was also examined for a number of other plant species. We chose a 69 nt sequence which includes exon 3 with 60 additional bp of flanking sequence (Figure S8A) to search 2.2 million pyrosequencing (454) ESTs from developing seeds of B. napus [14]. More than 200 of these ESTs were identified as WRI1 orthologs and of these > 80 spanned the exon 3 region. All of these reads included sequences that represent exon 3. Similarly, in the analysis of > 0.9 million castor ESTs [14], > 50 of these ESTs spanned the 5’ sequence region and all contained the VYL sequence. Finally, from analysis of > 4 million ESTs of oil palm mesocarp [15], > 200 ESTs spanned the exon 3 region, and all encoded ‘VYL”.

In addition, an analogous procedure was used to identify splicing at the 3’-end that distinguishes splice form 1 from splice forms 2 & 3. A search sequence of 32 nt located just before the last intron (spliced out in AtWRI1 splice form 1; Figure S9A), was used to identify B. napus ESTs that encoded the 3’ end (see Figure S9B). In the B. napus EST database, 80 ESTs were identified that closely matched the 3’ sequence of WRI1 and in all cases the stop codon position matched that of AtWRI1 splice forms 2 and 3, but not 1. The same conclusion was reached by analysis of 90 castor ESTs containing the 3’ stop codon. Taken together, for B. napus, castor and oil palm, we could not detect evidence for splice forms other than those corresponding to WRI1 splice form 3.

Monocot non-seed EgWRI1 is functional in dicot Arabidopsis plant

In addition to AtWRI1, the seed-expressed BnWRI1 or ZmWRI1 orthologs have been confirmed to function in regulation of seed oil biosynthesis by their ability to complement wri1 or to increase seed oil content [10,11,12]. Recently two AtWRI1 homologs (AtWRI3 and AtWRI4) that are stem and flower expressed, were shown to also activate expression of fatty acid biosynthetic genes and to complement the wri1 mutant [17]. In this study we asked whether a more highly diverged WRI1 ortholog associated with very high oil levels in non-seed tissue might have evolved different properties associated with this function. It was also unknown whether this shorter and divergent monocot non-seed EgWRI1 could function in regulating seed oil synthesis in a dicot or result in different phenotypes in germination and seedling development. Whether ZmWRI1a and ZmWRI1b can complement germination and seedling development phenotypes of wri1 has not been reported. Our work indicated that expression of EgWRI1 is able to rescue the ‘wrinkled’ seed coat (Figure 2), reduced seed oil content (Figure 3) and other phenotypes of wri1-1 (Figure 4). Interestingly, we noticed that the seed fatty profile of transgenic wri1-1 expressing EgWRI1 is not completely rescued compared to wri1-1 expressing AtWRI1 (Figure S1). Previous work also found that expression of ZmWRI1a or ZmWRI1b in a wri1 mutant did not restore the fatty acid profile to that of WT [11]. However, other than the phenotype of fatty acid profiles in transgenic wri1 plants, there were no differences found between transgenic wri1-1 plants expressing EgWRI1 or AtWRI1, in restoring wri1 mutant phenotypes to WT (Figures 2-4). These results indicate that despite major differences in their primary and secondary structure, this diverged EgWRI1 functions similarly to AtWRI1 in the wri1-1 mutant.

A question that is raised by this study concerns the function of the C-terminal regions of EgWRI1 and other WRI1 proteins, which are likely activation domains interacting with other factors of the transcription complex [26]. The highest sequence similarity between EgWRI1 and AtWRI1 is located in the N-terminus (~ 230 amino acids), which includes the crucial AP2 domains, a highly conserved feature in all AP2-type transcription factors. EgWRI1 is 93 amino acids shorter than AtWRI1 (Figure 1) and 92 of the amino acid ‘deletions’ are in the C-terminal region. After the AP2 domains, EgWRI1 is 111 amino acids long, compared to 203 amino acids for AtWRI1. In this study, we observed that fusion of an approximately 20 kDa protein tag (TAP) at the C-terminal region of EgWRI1 resulted in only minor restoration of the wri1-1 seed oil and wrinkled phenotypes (Figure S2-S4). However, EgWRI1-TAP retained some function, based on the fact that reduced germination of wri1 was still rescued (Figure S4A). In contrast, AtWRI1-TAP successfully complemented all phenotypes of wri1-1 that we examined.

Two predicted WRI1 splice forms are likely ‘artifacts’ of gene prediction software

Alternative splicing is an important form of regulation for many genes and can increase the functional diversity of transcripts in eukaryotic cells. Evidence presented in this manuscript indicates that splice form 3 of WRI1 is the only splice form that can be found in developing seeds, roots, flowers, and young seedlings of Arabidopsis and in developing seeds of B. napus and castor. Alternative splice forms in some cases produce protein products that lack or add functional domains, expanding their ability to interact within diverse cellular processes. For example, a splice form of Arabidopsis IDD14β encodes a product lacking a functional DNA-binding domain but plays a role in attenuating the activity of the full-length IDD14α through the formation of heterodimers [18]. Similarly, Arabidopsis CCA1 also produces a shortened alternative splice variant that results in the formation of nonfunctional heterodimers, decreased DNA binding activity and alters the circadian rhythm [22]. These and other examples, together with the alternative predicted forms of WRI1 raised the intriguing possibility that plant cells might use a similar self-regulatory mechanism of repressing or otherwise modifying WRI1 function through expressing alternative splice forms. However, all of our experimental evidence and bioinformatic searches failed to detect the presence of predicted WRI1 splice form 2 or 3.

The presence of WRI1 splice form 1 represented by one EST from suspension cultures treated with cycloheximide indicates that alternative splicing of WRI1 can occur. However, since alternative splice variants can be produced due to splice errors induced by stress [20], it is likely that cycloheximide led to aberrant splicing that is reflected by this one EST. Given the fact that WRI1 splice form 1 has been observed in only this one case and differs by only eight additional amino acids at the WRI1 C-terminal region (Figure 6B), we consider that a specific function of WRI1 splice form 1 under normal plant growth is very unlikely.

Taken together, we conclude that the WRI1 splice form 1 and 2, while predicted by automated annotation at the same confidence level as splice form 3, are unlikely to play a role in gene regulation, in the diverse tissues we examined. The prediction of forms 1 and 2 may be related to the fact that the sequences at the intron/exon junctions are not ‘cannonical’ which can lead to incorrect exon identification [27,28]. This ‘error’ in gene prediction is not specific to Arabidopsis because in the Phytozome database approximately one third of predicted protein sequences of WRI1 orthologs lack the VYL sequence (Figure S5B).

Small exon encoded “VYL” residues are a key component of AtWRI1 function

“VYL” is conserved in a number of WRI1 orthologs discovered in many plant species (Figure S5). The possible involvement of the “VYL” sequence of AtWRI1 in alternative splicing of WRI1 was discussed by Masaki et al., (2005), but biological or functional evidence to demonstrate a role of VYL in plants was not provided. Krizek (2003) used yeast to test the ability of a randomly mutagenized population of the Arabidopsis AP2-type transcription factor, AINTEGUMENTA (ANT) to bind to an ANT target sequence in yeast. Mutation of either “Y” or “L” lead to a full impairment of transcription activation while mutation of “V” lead to a reduced activation of ANT [29]. In the present study, we show that several mutated forms of residues “VYL” lead to failure or impaired ability to complement the low seed oil content of wri1-1 mutant (Figure 5) thus providing functional evidence in plants for the essential role of “VYL” for AtWRI1 function.

"VYL" is not a unique feature of WRI1-like transcription factors. The alignment of Arabidopsis AP2-type transcription factors indicates that all 18 members of the AP2 transcription factor family have a conserved protein sequence of "VYL" (http://planttfdb.cbi.pku.edu.cn/msa.php?sp=At&fam=AP2). Notably, amino acid “L” is more highly conserved and presumably most critical in AP2-type transcription factors. The results shown in Figure 5 supports this hypothesis by demonstrating that expression or HA-AtWRI1^L101Q in wri1-1 was the least effective in rescue of wri1-1 oil content, compared to HA-AtWRI1^V99D or HA-AtWRI1^Y100C (Figure 5).

In summary, this study has provided new insights into WRI1 structure at the protein and transcript level. The EgWRI1 sequence identified from oil palm mesocarp is highly similar to AtWRI1 over the N-terminal 230 amino acids, but surprisingly divergent in sequence, in length and in predicted secondary structure over the remainder of the protein’s C-terminal domain. Nevertheless, the EgWRI1 protein is functional in restoring oil content, germination and seedling establishment of the Arabidopsis wri1-1 mutant, implying that the C-terminus may have a more general role in maintaining WRI1 structure/function, rather than, for example, interactions with specific DNA sequences. Second, the conserved VYL small exon of WRI1 was shown to be essential for full WRI1 function. Mutations of all three residues caused complete loss of ability to complement wri1-1, and single mutations at the “L” residue were more negative than at V or Y. Finally, we have established that splice form At3g54320.3 is the only one of the three predicted splice forms that is expressed under normal growth conditions of seeds, roots, flowers and seedlings of Arabidopsis.

Plant Materials

Arabidopsis (Arabidopsis thaliana) wild-type (Columbia-2 ecotype) and wri1-1 [1] were used in this study. Plants were grown in a growth chamber on potting mix at 22°C with a 16 h light (100-150 µmol m⁻² s⁻¹ illumination)/8 h dark photoperiod cycle. For experiments with plants grown on plates, seeds were surface sterilized in 70% (v/v) ethanol (containing 0.05% (v/v) Tween 20), following by rinsing in 95% (v/v) and pure ethanol. Sterilized seeds were spread on petri dishes containing half-strength Murashige and Skoog (MS) medium (Caisson), 2.6 mm MES (pH 5.7; adjusted with KOH), 1% sucrose (unless noted otherwise in the figure legends), and 0.8% agar. Seeds were stratified at 4°C in the darkness for 2-3 d prior to use.

Plasmid Construction and Arabidopsis Transformation

The oil palm EgWRI1 gene was synthesized by GeneArt based on cDNA sequences obtained from oil palm mesocarp [15]. The nucleotide sequence of synthetic EgWRI1 is in Figure S10. Site-directed mutations (AtWRI1^{V99A/Y100A/L101A}; AtWRI1^V99D; AtWRI1^Y100C; AtWRI1^L101Q) were introduced into the AtWRI1 coding sequence (CDS) by PCR (see Table S1). The modified genes were subcloned into binary vectors pEarleyGate 201 or pEarleyGate 205 [30]. Constructs were introduced into wri1-1 mutants through Agrobacterium tumefaciens (GV3101 strain)-mediated transformation by floral dipping [31]. Transgenic seedlings were selected with 10 µg/mL Basta (Sigma-Aldrich) on plates. Genomic DNA of transgenic seedlings was extracted and gene insertion was confirmed by PCR using a 35S promoter forward primer and a gene-specific reverse primer. Homozygous plants were used in all experiments except T2 transgenic plants in experiments presented in Figure 5.

Scanning electron microscopy

Sample preparation followed methods previously described [2]. Samples were examined in a JEOL 6610LV SEM (tungsten hairpin electron emitter) scanning electron microscope (JEOL Ltd.).

Germination and Seedling Establishment Assays

Surface-sterilized seeds were spread on half-strength MS plates (containing 1% sucrose). Seeds were stratified for 2-3 d prior to being transferred to a growth chamber. Germination was scored by radicle emergence 2 d after imbibition. To determine seedling establishment, plants were grown vertically on half-strength MS plates (without the addition of sucrose) and 10-day-old seedlings were evaluated for seedling establishment. Alternatively, seeds were grown in potting mix and two-week-old seedlings were used to count the number of the seedlings that had established.

RT-PCR for WRI1 transcript amplification and analysis

Whole seedlings were harvested from plates, ground in liquid nitrogen, and total RNA was isolated using the RNeasy Plant Mini kits (Qiagen). Genomic DNA contamination was removed using DNase I (Qiagen). First-strand cDNA was synthesized using the Reverse Transcription System (Promega). Genomic DNA contamination was not found in the RNA samples treated with DNase I. PCR products were purified using the Gel and PCR Clean-Up System (Promega).

RNASeq analysis of WRI1 transcripts from Arabidopsis

Developing seeds were collected from liquid nitrogen frozen siliques at 7-8, 9-10, and 11-12 d after flowering. Siliques were opened over dry ice and frozen seeds were separated from the silique walls by filtering through a liquid nitrogen cooled sieve into a tube on dry ice. Approximately 100 mg of developing seeds was finely ground and RNA extracted as described [32] and analyzed for yield and quality by capillary electrophoresis (Agilent 2100). Libraries for sequencing were prepared from 2–4 µg total RNA using Illumina TruSeq RNA kits and sequenced with Illumina HiSeq2000. Reads (50 nt) were trimmed, filtered and aligned to TAIR10 using TopHat v 1.4.1 (Parameters: --no-novel-juncs; -G TAIR10.gff) and Bow tie v 0.12.7. Cufflinks v 2.0.2 was used to generate gene FPKM expression measures. The results were also loaded into a genome browser for inspection. Three different read mappings were performed to assess WRI1 alternative splice forms. Samples were aligned to TAIR10 using the CLC Genomics, Map Reads to Reference Tool and Large Gap Read Mapping Tool (Parameters: Mismatch cost 2; Insertion cost 3; Deletion cost 3; Similarity 0.9; Length fraction 0.9). 117 million RNASeq reads for roots (SRR331219,SRR331224) and 58.4 million for flowers (SRR388668, SRR388670, SRR013413, SRR013416) were downloaded from the NCBI short read archive and mapped to TAIR10 as described above.

Fatty Acid Analysis

Arabidopsis seed oil content analysis followed the method described previously [33], with minor modification. In brief, twenty Arabidopsis seeds were transesterified directly in a glass tube by addition of 1 mL freshly prepared sulfuric acid in methanol (5% (v/v)), 25 µL of BHT solution (0.2% butylated hydroxy toluene in methanol), 25 µg of triheptadecanoin (as internal standard) and 300 µL of toluene. After reaction at 90°C for 90 min the fatty acid methyl ester extracts were extracted and analyzed by gas chromatography with a DB23 column.

Statistical Analysis

A binomial test using R programming was used to calculate the statistical significance of conclusions on AtWRI1 splice form abundance. The student’s t-test was performed to evaluate the statistical confidence in differences observed between controls and samples expressing different WRI1 constructs.