Transposable Element Insertion and Epigenetic Modification Cause the Multiallelic Variation in the Expression of FAE1 in Sinapis alba.

Using the FATTY ACID ELONGATION1 (FAE1) gene as a case study, this work reveals that transposable element insertion and epigenetic modification generate multiple heritable alleles varying in expression level and modulating trait phenotype from qualitative to quantitative variation. Naturally occurring heritable variation provides a fundamental resource to reveal the genetic and molecular bases of traits in forward genetic studies. Here, we report the molecular basis of the differences in the four alleles E1, E2, E3, and e of the FATTY ACID ELONGATION1 (FAE1) gene controlling high, medium, low, and zero erucic content in yellow mustard (Sinapis alba). E1 represents a fully functional allele with a coding DNA sequence (CDS) of 1521 bp and a promoter adjacent to the CDS. The null allele e resulted from an insertional disruption in the CDS by Sal-PIF, a 3100-bp PIF/Harbinger-like DNA transposon, whereas E2 and E3 originated from the insertion of Sal-T1, a 4863-bp Copia-like retrotransposon, in the 5′ untranslated region. E3 was identical to E2 but showed cytosine methylation in the promoter region and was thus an epiallele having a further reduction in expression. The coding regions of E2 and E3 also contained five single-nucleotide polymorphisms (SNPs) not present in E1, but expression studies in Saccharomyces cerevisiae indicated that these SNPs did not affect enzyme functionality. These results demonstrate a comprehensive molecular framework for the interplay of transposon insertion, SNP/indel mutation, and epigenetic modification influencing the broad range of natural genetic variation in plants.


INTRODUCTION
Genetic and molecular mechanisms of many traits in plants have been revealed based on naturally occurring heritable variation. Allelic differences in nucleotide sequences resulting from singlenucleotide polymorphisms (SNPs), indels (insertions or deletions), and transposable element (TE) insertions are primarily responsible for contrasting trait phenotypes. TEs represent a large proportion of the nuclear genomes in many plant species (Vicient et al., 1999;Meyers et al., 2001). They are classified into class I (RNA) and II (DNA) based on their different modes of mobility (Charlesworth et al., 1994). Class I elements are retroelements that use reverse transcriptase to transpose via an RNA intermediate, whereas class II elements can transpose directly from DNA to DNA. TEs and the movement of TEs provide an important source of new genetic variation in both plants and animals (Kidwell and Lisch, 1997). Insertions of TEs in exons, introns, and 59 flanking sequences of genes have resulted in new alleles having no or reduced function. Allelic variation of the knotted1 locus resulted from the insertion of Mu1 or Mu8 within a 310-bp region of the kn1 third intron in maize (Zea mays) (Greene et al., 1994). In the blood orange (Citrus sinensis), the expression of Ruby (regulator of anthocyanin biosynthesis) was found to be modified by the expression of a Copia-like retrotransposon inserted in its upstream region (Butelli et al., 2012).
Epigenetic modifications including DNA cytosine methylation, histone variants, and small RNAs provide another molecular mechanism for regulating gene expression and generate a potential source of natural heritable variation (for reviews, see Kalisz and Purugganan, 2004;Richards, 2006Richards, , 2011Becker and Weigel, 2012;Weigel and Colot, 2012). The clark kent (clk) epiallele described in Arabidopsis thaliana contains hypermethylated cytosines in the coding region of the SUPERMAN gene (Jacobsen and Meyerowitz, 1997). The clk epiallele showed a phenotype similar to loss-of-function mutants of SUPERMAN, having increased numbers of stamens and carpels. Genes controlling ecologically important traits such as floral symmetry in toadflax (Linaria vulgaris) and pathogen resistance in Arabidopsis show differences in expression due to differential methylation (Cubas et al., 1999;Stokes et al., 2002). All natural epialleles reported to date have involved a gain or loss of DNA methylation (for a review, see Weigel and Colot, 2012).
Erucic acid (cis-13 docosenoic acid [C22:1]) is one of the major fatty acids produced in the seeds of many plant species, including Brassica species and yellow mustard (Sinapis alba, SS, 2n=24). High erucic oil and erucic acid derivatives have many potential industrial applications, including as lubricants, surface coatings, and pharmaceuticals (Sonntag, 1991(Sonntag, , 1995Leonard, 1994). However, erucic acid is undesirable in edible oil due to its antinutritional effects (Sauer and Kramer, 1983). The biosynthesis of erucic acid occurs in the cytoplasm via an elongation complex containing a seed-specific condensing enzyme, b-ketoacyl-CoA synthase (KCS) (Stumpf and Pollard, 1983). KCS is encoded by the FATTY ACID ELONGATION1 (FAE1) gene, which has been cloned from Arabidopsis (James et al., 1995), jojoba (Simmondsia chinensis) (Lassner et al., 1996), turnip rape (Brassica rapa), cabbage (Brassica olereacea) (Das et al., 2002), oilseed rape (Brassica napus) (Han et al., 2001), and many other plant species. Additionally, FAE1 genes from a number of high and low erucic acid cultivars of B. napus have been cloned and sequenced. In the FAE1 gene from the A genome of B. napus, a single base change of nucleotide 845 from cytosine to thymine resulted in a loss-of-function allele (Han et al., 2001;Roscoe et al., 2001;Katavic et al., 2002;Wu et al., 2007Wu et al., , 2008. A second mutation consisting of a 4-bp deletion between T1366 and G1369 in the FAE1 gene, which led to a frameshift mutation, was predominantly found in the C genome (Wu et al., 2007(Wu et al., , 2008. Thus, the known erucic acid variants in B. napus have resulted from SNP/indel mutations of FAE1. Erucic acid variants have been identified by inbreeding of open-pollinated plants in yellow mustard, which is an obligate outcrossing crop due to its self-incompatible reproductive system (Raney et al., 1999;Cheng et al., 2012). The high (52.9%), medium (24.4%), low (1.4%), and zero (0.1%) erucic contents of the yellow mustard lines Y517, Y496, Y1130, and Y514 were conditioned by the four FAE1 alleles E 1 , E 2 , E 3 , and e, respectively (Javidfar and Cheng, 2013). Here, we describe the TE insertion and epigenetic variation in FAE1 that lead to the observed changes in erucic acid content in yellow mustard.

RESULTS
The Four FAE1 Alleles E 1 , E 2 , E 3 , and e Show Variation in Coding DNA Sequence The coding regions of E 1 , E 2 , and E 3 were successfully cloned from lines Y517 (E 1 E 1 ), Y496 (E 2 E 2 ), and Y1130 (E 3 E 3 ), respectively, using primer pair No 1 (Supplemental Table 1; Figure  1A). E 1 had a coding DNA sequence (CDS) of 1521 bp encoding a polypeptide of 507 amino acids, which contained six conserved cysteine residues, four conserved histidine residues, and a conserved serine residue compared with the polypeptides encoded by the FAE1 genes in Arabidopsis, B. napus, Brassica juncea, B. rapa, and B. oleracea (Supplemental Figure 1). The sequences of FAE1 alleles E 2 and E 3 were identical in the coding region. Sequence alignment with E 1 allowed the identification of five point mutations at positions 788, 792, 903, 1035, and 1176 in the CDS of E 2 and E 3 ( Figure 1A). Only the mutation at position 788 (A of E 1 to G of E 2 and E 3 ) led to an amino acid change; the arginine residue at position 263 in the protein sequence of E 1 was substituted by lysine in the protein encoded by the E 2 and E 3 alleles (Supplemental Figure 1). However, primer pair No 1 failed to amplify the CDS of the FAE1 allele e of line Y514 (ee). A 1328-bp fragment from nucleotide 194 to 1521 was successfully obtained for the allele e using primer pair No 2 (Supplemental Table 1; Figure 1A). Sequence alignment with E 1 indicated that eight point mutations occurred at positions 439, 532, 601, 922, 1040, 1097, 1295, and 1470 in the partial CDS of e, but none of these led to amino acid changes. An additional 1447-bp DNA fragment was cloned from allele e by PCR walking using primer pair No 3 (Supplemental Table 1 and Supplemental Figure 2C). This fragment included another 32 bp from the CDS (nucleotide 162 to 193) and an unknown DNA fragment of 1415 bp. Thus, although 1360 bp of the CDS from nucleotide 162 to 1521 for the FAE1 allele e was cloned, the remaining 161 bp CDS from nucleotide 1 to 161 was not obtainable using these primer sets. BLAST analysis of the unknown 1415-bp fragment against the sequence database of Arabidopsis (TAIR) revealed that it had high similarity to a transposable element (At3G15310, E value = 9e 285 ). Alignment with the E 1 allele indicated that the transposable element was inserted into the CDS of e at position 162 ( Figure 1A). Therefore, it was inferred that the insertion of the 1415 bp transposable element in the coding region of FAE1 led to the occurrence of the loss-of-function allele e. However, the use of these primer sets did not allow us to determine whether the entire sequence of the transposable element had been obtained.
Heterologous Expression of FAE1 in Yeast Indicates That Enzymes Encoded by Alleles E 1 and E 2 Are Functional and Have Similar Expression Level Transgenic yeast (Saccharomyces cerevisiae) cultures containing the empty construct pYES2.1/V5-His-TOPO produced the typical fatty acids, i.e., C16:0, C16:1, C8:0, and C18:1, found in untransformed yeast cells ( Figure 1B). Transgenic yeast cells harboring the yeast expression vector pYES2.1/V5-His-TOPO-E 1 carrying E 1 coding sequence or pYES2.1/V5-His-TOPO-E 2 with E 2 coding sequence produced the novel long-chain fatty acids C20:1, C22:1, and 24:0, and erucic acid content was not significantly different between cultures expressing E 1 versus E 2 ( Figure 1B). These data indicated that the cloned alleles E 1 and E 2 encoded functional enzymes capable of elongation of C18:1 to C20:1 and C22:1 and had similar activity levels. Therefore, the reduction of erucic acid content in the yellow mustard lines Y496 and Y1130 did not appear to be caused by the nucleotide mutations in the CDS of E 2 and E 3 .
The Three FAE1 Alleles E 1 , E 2 , and E 3 Exhibit Significant Differences in Transcript Levels Transcripts of the FAE1 were only detected in 14-to 35-d-old embryos and were absent in leaf, stem, flower bud, and open flower tissue of Y517, Y496, and Y1130 (Supplemental Figure  3A), suggesting that E 1 , E 2 , and E 3 were specifically expressed in the embryo. Transcripts were not detected in immature embryo tissue of Y514 since Y514 carried the loss-of-function FAE1 allele e (Supplemental Figure 3A).
Peak transcript levels of E 1 , E 2 , and E 3 were observed in 21-dold embryos ( Figure 1C). E 1 had the highest expression level ( Figure 1C), and the expression levels of E 2 and E 3 were 21.6% (SD 6 2.3%) and 1.1% (SD 6 0.1%) of that of E 1 , respectively. E 3 had the lowest expression level, at 4.8% (SD 6 0.1%) of E 2 . The transcript levels of E 1 , E 2 , and E 3 were positively correlated with erucic acid contents in the seeds of Y517, Y496, and Y1130. Thus, although heterologous expression in yeast indicated that the expressed proteins encoded by E 1 , E 2 , and E 3 had the same activity level in the elongation of C18:1 to C20:1 and C22:1, significant differences at the transcript level were detected among these alleles by quantitative PCR (qPCR).
Histochemical GUS Assays Indicate That the 877-bp 59 Flanking Fragment of E 1 Functions as a Seed-Specific Promoter DNA fragments comprising the 300 bp downstream from the translation stop codon were cloned for E 1 , E 2 , and E 3 using primer pair No 4 (Supplemental Table 1 and Supplemental Figure 2) and had the same nucleotide sequence, indicating that they were not related to the reduction of transcript levels seen for E 2 and E 3 . To determine if the 59 upstream (promoter) region was involved in the differential transcription of the FAE1 alleles, the 59 flanking sequences of E 1 , E 2 , and E 3 were cloned by PCR walking using primer pair No 5 (Supplemental Table 1 and Supplemental Figure 2). The 1095-bp 59 flanking fragment cloned for E 1 contained a 703-bp putative promoter, which showed 87% identity to that of the FAE1 promoter region from the high erucic acid B. napus line Askari. The 1221-bp 59 flanking fragments cloned for E 2 and E 3 shared the same nucleotide sequence, but did not exhibit any similarity with the putative promoter sequence of E 1 . (A) Schematic representation of the structure of the four FAE1 alleles E 1 , E 2 , E 3 , and e. Black rectangle, the entire coding region; gray rectangle, 300-bp 39 UTR; solid black arrow, promoter; vertical long arrow, point mutation; inverted triangle, insertion; asterisk, putative transcription start site at position 24935; red vertical long arrow, cytosine methylation at positions 25146, 25095, and 24943; dashed lines, deletion; OF/OR, primer set No 1 for cloning the entire coding region; PF/OR, Primer set No 2 for cloning the partial coding region of e; WF/WR, primer set No 14 for cloning the entire FAE1 gene. (B) Gas chromatography analysis of fatty acid composition of yeast cells containing the construct pYES2.1/V5-His-TOPO-E 1 with the FAE1 allele E 1 (black), the construct pYES2.1/V5-His-TOPO-E 2 with the FAE1 allele E 2 (white), and the empty vector pYES2.1/V5-His-TOPO (gray color, control). (C) Real-time qPCR analysis of expression levels of the FAE1 alleles E 1 , E 2 , and E 3 in 14-to 35-d-old embryos. Error bars indicated SD of the mean. Actin2 (FG576123) was used as an internal control. To test if the isolated 59 flanking fragments included functional promoters, the 877-bp region upstream from the translational start codon of E 1 and the 1003-bp upstream region of E 2 and E 3 were cloned using primer pairs No 16 and 17, respectively (Supplemental Table 1), then placed in front of a b-glucuronidase (GUS) reporter gene in a plant expression vector. The resulting constructs pBI101-Pro E1 -GUS and pBI101-Pro E2 -GUS were transformed into Arabidopsis. In transgenic plants containing the construct pBI101-pro E1 -GUS, histochemical assays led to the detection of GUS activity in the embryo ( Figure 1D), but not in the leaf, stem, bud, or flower tissues (Supplemental Figures 3B to 3E), indicating that the 877-bp 59 flanking sequence of E 1 functioned as a seed-specific promoter. In contrast, no GUS activity was detected in the embryo or any vegetative tissues of the transgenic plants containing the construct pBI101-pro E2 -GUS, suggesting that the 1003-bp 59 flanking sequence of E 2 and E 3 did not function as a promoter.
Insertion of a Copia-Like Retrotransposon in the 59 Untranslated Region Separates the FAE1 Promoter from the CDS in Alleles E 2 and E 3 To determine if E 2 and E 3 had promoters that were located further away from the coding region, additional upstream regions were cloned by PCR walking. Four rounds of PCR walking using primer pairs No 6, 7, 8, and 9 (Supplemental Table 1 and Supplemental Figure 2B) allowed the isolation of four additional 59 upstream fragments of 673, 741, 791, and 903 bp, respectively, from both E 2 and E 3 . Sequence alignments indicated that the 3337-bp 59 flanking fragments of E 2 and E 3 obtained using primer pairs No 5-9 shared the same nucleotide sequence but did not exhibit any similarity with the promoter sequence of E 1 . However, the promoters of E 2 and E 3 were successfully cloned using primer pair No 10, which was designed based on the promoter sequence of E 1 (Supplemental Table 1 and Supplemental Figure 2). The E 2 and E 3 promoters had the same nucleotide sequence and length (703 bp) as the promoter of E 1 . The gap between the E 2 /E 3 promoter region and the 3337-bp 59 flanking fragment was filled with a 1717-bp sequence obtained using primer pair No 11 (Supplemental Table 1 and Supplemental Figure 2B). Sequence alignments of the promoter and the 1717-and 3337-bp 59 flanking fragments revealed an insertion of 4863 bp between the coding region and promoter in both alleles. Sequence alignment with E 1 indicated that the 4863-bp insertion was located in the 59 untranslated region (UTR) of E 2 and E 3 at position 236 ( Figure 1A). An additional 600 bp upstream of the promoter region was obtained for E 1 and E 2 /E 3 using primer pair No 12 (Supplemental Table 1 and Supplemental Figure 2B), and the sequence of this region was the same in these alleles.
The 4863-bp insertion contained a primer binding site (PBS) and a polypurine tract (PPT), with 189-bp long terminal repeats (LTRs) on each side ( Figure 2A; Supplemental Figure 4). It was flanked by target site duplication sequence ( Figure 2A). These characteristics implied that the inserted 4863-bp DNA fragment was a retrotransposon. The protein encoded by this retrotransposon had several conserved domains including regions with sequence similarity to retrotransposon gag, reverse transcriptase, and the Ty1/Copia family of RNase H (Supplemental Figure 5) and exhibited 58% similarity to a putative retroelement pol polyprotein in Arabidopsis. Therefore, the 4863-bp insertion in the 59 UTR of E 2 and E 3 appears to be a Copia-like retrotransposon and was designated as Sal-T1. The Sal-T1 is oriented in the antisense transcriptional direction (Figure 2A). The two LTRs of Sal-T1 had the same nucleotide sequence, suggesting that the insertion of Sal-T1 in the FAE1 could be a relatively recent event. The increased physical distance between the promoter and the CDS due to the insertion of Sal-T1 in the 59 UTR might lead to the observed reduction in mRNA transcription of E 2 and E 3 alleles, thereby resulting in the decrease of erucic acid content seen in lines Y496 and Y1130.
A 3100-bp PIF/Harbinger-Like DNA Transposon Is Inserted into the CDS of FAE1 Allele e The promoter of the FAE1 allele e was cloned using primer pair No 10 (Supplemental Table 1 and Supplemental Figure 2). It had a size of 446 bp due to a 7-bp deletion at position 2350 and a 250-bp deletion at position 2410 compared with the promoter sequences of E 1 , E 2 , and E 3 ( Figure 1A). The gap between the promoter and the 1415-bp transposable element inserted in the CDS of e was bridged by a 2082-bp DNA fragment obtained using primer pair No 13 (Supplemental Table 1 and Supplemental Figure 2C). The 2082-bp fragment contained the remaining 161 bp (nucleotide 1 to 161) of the coding region and an additional 1685 bp of transposable element sequence. In summary, the mutant allele e had a 1521-bp CDS with a 3100-bp transposable element inserted at position 162 and a promoter adjacent to the start codon of the CDS ( Figure 1A).
The 3100-bp insertion had a 13-bp terminal inverted repeat on each end ( Figure 2B; Supplemental Figure 6) and was flanked by a target site duplication sequence TAA. These characteristics suggested that the 3100-bp insertion was a DNA transposon. The insertion contained two open reading frames (ORF1 and ORF2) ( Figure 2B; Supplemental Figure 6). The protein encoded by ORF1 was 103 amino acids in length (Supplemental Figure  7A) and did not have any similarity with amino acid sequences in the National Center for Biotechnology Information protein database. The protein encoded by ORF2 was 383 amino acids in length (Supplemental Figure 7B). It had a putative Asp-Asp-Glu (DDE) motif (referred to as N2, N3, and C1) similar to that found in the PIF element of maize (Zhang et al., 2001) ( Figure 2B). ORF2 encoded a domain with high similarity (E = 3.87e 2129 ) to the transposases of the PIF/Harbinger superfamily. This suggested that the inserted DNA fragment in the coding region of the FAE1 allele e was a DNA element putatively encoding a transposase belonging to the PIF/Harbinger superfamily, and it was therefore designated as Sal-PIF.
Thus, the entire nucleotide sequence for each of the four FAE1 alleles E 1 , E 2 , E 3 , and e with sizes of 2524, 7424, 7424, and 5369 bp, respectively, was obtained by PCR walking. To validate these results, the four FAE1 alleles were each cloned in their entirety using primer pair No 14 (Supplemental Table 1; Figure 1A). As expected, the resulting DNA fragments of the four FAE1 alleles had the same sizes (Supplemental Figure 8) and nucleotide sequences as those obtained from PCR walking.

The FAE1 Allele E 3 Is an Epiallele
The FAE1 alleles E 2 and E 3 had the same coding and promoter sequences and contained a Copia-like retrotransposon Sal-T1 in the 59 UTR, but exhibited significant differences in transcription level ( Figure 3B), and seeds of plants carrying these alleles showed significant differences in erucic acid content ( Figure 3A). To determine if DNA methylation was involved in regulating the expression of E 3 , we analyzed the methylation status of the coding and promoter regions of the three alleles E 1 , E 2 , and E 3 in 21-d-old embryos from different plants for each of the three lines Y517, Y496, and Y1130 using a bisulfite sequencing procedure. None of the three alleles were methylated in the coding regions (nucleotides 1 to 1515) (Supplemental Figure 10). As expected, the allele E 1 of line Y517 did not show methylation in its promoter region (nucleotides 2335 to 1) ( Figure 3C; Supplemental Figure 11) nor was DNA methylation detected in the promoter region (nucleotides 25234 to 24900) of E 2 ( Figure 3D; Supplemental Figure 12). However, CG methylation was detected at positions 25146, 25095, and 24943 in the promoter region (nucleotides 25234 to 24900) of the allele E 3 ( Figure 3E; Supplemental Figure 13). These data suggested that the FAE1 allele E 3 is an epiallele.
Twenty-five plants were obtained from the 45 5-azaC-treated seeds. Of the 25 plants, 13 (52%) plants were revertants induced by 5-azaC treatment and produced seeds with significantly higher erucic content (average, 7.6%; range, 4.3 to 15.4%) than the control plants of Y1130 (average, 1.6%; range, 0.7 to 3.4%) (t = 13.8, P = 0) ( Figure 3A). The expression level of E 3 in these revertants was also higher than that of E 3 in the control plants of Y1130 ( Figure 3B). DNA methylation analysis of the revertants indicated that the FAE1 allele E 3 was demethylated at positions 25146 and 25095 but remained methylated at position 24943 ( Figure 3F; Supplemental Figure 14). These results confirmed that E 3 is an epiallele and the methylation in the promoter region of the FAE1 allele E 3 led to the reduction in transcription, thereby decreasing the erucic acid content.
Furthermore, 100 untreated progeny seeds of Y1130 were analyzed for erucic content using the half-seed technique (Downey and Harvey, 1963). Ninety-nine seeds had low erucic content (range: 0.7 to 3.4%), whereas one seed (Y1130-16) contained 12.7% erucic acid and was a spontaneous revertant. The plant germinated from the seed Y1130-16 produced seeds with an average erucic content of 12.3% (range: 11.6 to 12.7%). Analysis by qPCR indicated that the expression level of E 3 in this revertant was higher than that of E 3 in Y1130 (control) (Figure 3B). DNA methylation analysis of this plant showed that the FAE1 allele E 3 was also demethylated at positions 25146 and 25095, but remained methylated at position 24943 ( Figure 3G; Supplemental Figure 15), as did those revertants induced by 5-azaC. These results indicated the metastability of the epiallele E 3 . Thus, we concluded that the FAE1 allele E 3 is an epiallele and designated it as E epi3 to differentiate it from the nonepialleles E 1 , E 2 , and e.
The Allele-Specific Markers for e and E 2 Cosegregate with the Zero and Medium Erucic Acid Contents in F2 Populations of Y514 (ee) 3 Y517 (E 1 E 1 ) and Y496 (E 2 E 2 ) 3 Y517 (E 1 E 1 ), Respectively Primer pair No 31 (Supplemental Table 1) generated one dominant marker of 1616 bp specific for the e locus carrying the transposon Sal-PIF. Cosegregation of the dominant marker with the zero erucic acid trait was studied in 96 F2 plants from the cross Y514 (ee) 3 Y517 (E 1 E 1 ). The dominant marker was present in all F2 plants with erucic acid content (range: 0.0 to 0.2%) resembling the zero erucic acid parent Y514 (0.0 to 0.1%), but absent in the F2 plants with high erucic content (range: 39.5 to 58.0%) resembling levels found in Y517 (range: 50.3 to 55.5%) ( Figure 4A). The cosegregation of this marker with the zero erucic trait suggested that the loss-of-function FAE1 allele e was due to the insertion of the DNA transposon Sal-PIF in its coding region.
Primer pair No 32 (Supplemental Table 1) produced a dominant marker of 768 bp specific for the region carrying the retrotransposon Sal-T1 in the 59 UTR of E 2 . Cosegregation of the dominant marker with the medium erucic content was studied in 70 F2 plants of the cross of Y496 (E 2 E 2 ) 3 Y517 (E 1 E 1 ). The dominant marker was present in all the F2 plants with erucic acid content (range: 20.0 to 29.8%) similar to that of the medium erucic acid parent Y496 (21.7 to 27.1%), but absent in the F2 plants with high erucic content (range: 44.4 to 49.2%) similar to that of Y517 (50.3 to 55.5%) ( Figure 4B), implying that the reduction of erucic content in Y496 was due to the Sal-T1 insertion in the 59 UTR of E 2 .

DISCUSSION
In this study, we investigated the molecular and epigenetic mechanisms leading to the occurrence of four different alleles of the FAE1 in yellow mustard. The FAE1 allele E 1 conditioning high erucic content had a coding sequence of 1521 bp, which  was in good agreement with lengths of the FAE1s of B. napus and Arabidopsis (James et al., 1995;Barret et al., 1998;Fourmann et al., 1998). The predicted polypeptide of 507 amino acids encoded by the FAE1 allele E 1 also showed high similarity with the proteins encoded by the FAE1s in B. juncea (97% identity), B. napus (96% identity), B. rapa (96% identity), B. oleracea (96% identity), and Arabidopsis (86% identity) (Supplemental Figure 1). Four conserved histidine residues in the polypeptide (Supplemental Figure 1) were demonstrated to be essential for the function of the FAE1 in Arabidopsis (Ghanevati and Jaworski, 2001), while a conserved serine residue at position 282 (Supplemental Figure 1) was shown to be essential for the FAE1 functionality in B. napus (Han et al., 2001;Katavic et al., 2002). The promoter of E 1 was 703 bp and proved to be seed specific and functional. It had high similarity to the promoter of the wild-type FAE1 in Brassica species (Han et al., 2001). Yellow mustard is phylogenetically related to Brassica and Arabidopsis species (Warwick and Black, 1991), and the high level of similarity between the FAE1 genes of these species likely reflects their common evolutionary origin.
This report reveals that transposable element insertion and epigenetic modification are involved in modulating the expression of FAE1 in the erucic acid-containing species of Brassicaceae. In B. napus, SNP/indel mutations in the coding region of the FAE1 resulted in loss-of-function alleles (Han et al., 2001;Roscoe et al., 2001;Katavic et al., 2002;Wu et al., 2007Wu et al., , 2008. In contrast, the loss-of-function FAE1 allele e in yellow mustard was caused by an insertional disruption in its coding region by Sal-PIF, a PIF/ Harbinger-like DNA transposon. The FAE1 allele E 2 , controlling the medium erucic acid trait, originated from the insertion of a Copia-like retrotransposon Sal-T1 in its 59 UTR. The low erucic allele E 3 was identical to E 2 in nucleotide sequence but was shown to be an epiallele due to DNA methylation in its promoter region. The different molecular mechanisms underpinning the allelic variation of FAE1 in yellow mustard and Brassica species suggested that the mutations of FAE1 occurred after the divergence of the species. Insertions of DNA transposons or retrotransposons within or near genes have been reported to negatively affect the expression of genes by decreasing or abolishing transcription (Kumar and Bennetzen, 1999;Hollister and Gaut, 2009;Ahmed et al., 2011;Eichten et al., 2012Eichten et al., , 2013. For example, the null wx-ml allele originated from the insertion of the 409-bp DNA transposon Dissociation 1 in the 9th exon of the waxy (wx) gene in maize (Wessler, 1991). The gypsy-like retrotransposon Gret1 inserted upstream of the mybA1 coding sequence blocked gene expression, causing the loss of pigmentation in white cultivars of grape (Vitis vinifera) (Kobayashi et al., 2004;Fournier-Level et al., 2010). Here, our data further confirmed that mutations induced by TE insertions can have negative effects on gene expression. The insertion of Sal-PIF in the coding region of the FAE1 caused the loss-of-function FAE1 allele e. The insertion of the Copia-like retrotransposon Sal-T1 in the 59 UTR of E 2 and E 3 resulted in reduced transcription levels. There are three possible explanations for this: First, the promoters of E 2 and E 3 are farther from the CDS due to the insertion of Sal-T1, which might result in a reduction in transcription efficiency compared with E 1 , where the promoter is adjacent to the coding region. Second, the inserted Sal-T1 might disturb the interaction of the promoter and other transcriptional factors, thereby affecting FAE1 expression. Third, the transcription direction of Sal-T1 was opposite to that of E 2 and E 3 . Therefore, transcripts of E 2 and E 3 might form senseantisense pairs by pairing with the transcript of Sal-T1 on the opposite strand, causing a modulation of gene expression. Jacobsen and Meyerowitz (1997) reported that the clk epialleles were extensively methylated, covering the start of transcription and most of the transcribed region in Arabidopsis. In tomato (Solanum lycopersicum), the colorless nonripening (Cnr) epimutant exhibited high levels of cytosine methylations in the SPL-CNR promoter region (Manning et al., 2006). However, our study revealed that the expression levels of the FAE1 epiallele E 3 were associated with the presence/absence of only three methylated cytosines in the promoter region. This could be due to that the methylated cytosines are critical for the FAE1 promoter activity. Smale and Kadonaga (2003) reported that transcription initiation of genes generally relies on a simple core promoter sequence that can extend ;35 bp upstream and/or downstream of the transcription initiation site in eukaryotes. The four nucleotides "ACGT" in the promoter can act as the transcription factor recognition site (Jakoby et al., 2002). In the promoter of FAE1 allele E 3 , the four nucleotides "ACGT" were positioned from 25145 to 25148 and the cytosine at position 25146 was methylated. In addition, methylation was observed at position -4943 C, which is very close to the predicted transcription start site (nucleotide 24935) in the promoter of E epi3 ( Figure 1A) (http://www.fruitfly.org/seq_tools/promoter.html). Methylation at these critical positions in the promoter of E epi3 could have a negative effect on binding of the transcription preinitiation complex and therefore result in reduction of transcription efficiency and lowered erucic acid content. The revertants of Y1130 had increased erucic acid content due to demethylation at positions 25146 and 25095, confirming that the methylation level in the promoter of E epi3 is negatively correlated with the erucic content in the seed. Interestingly, the methylated cytosine at position 24943 was close to the retrotransposable element Sal-T1 (nucleotide 24899 to 237), and this position was not demethylated either by 5-azaC treatment or spontaneously. It has been reported that DNA methylation on and near transposons is highly stable, whereas it is less stable within genes at a distance from transposons (Weigel and Colot, 2012). The genetic and molecular mechanisms underlying the methylation status of E 3 remains to be studied.
Our previous genetic studies indicated that E 3 was heritable and acted in an additive manner with E 2 , but fit to a partial dominance model with E 1 in yellow mustard (Javidfar and Cheng, 2013). The present molecular analysis revealed that E 3 is an epiallele differing from E 2 due to DNA methylation in its promoter region. Heritable epialleles have been reported in many plant species such as Arabidopsis (Jacobsen and Meyerowitz, 1997), tomato (Messeguer et al., 1991), and toadflax (Cubas et al., 1999). Recently, genome-wide DNA methylation analyses have further confirmed the occurrence of epigenome-wide inheritance of methylation variants in soybean (Glycine max; Schmitz et al., 2013) and maize (Regulski et al., 2013). However, it is well documented that epialleles are not stable and occasionally revert phenotypically. The clk-3 mutant spontaneously reverted to wild-type or nearly wild-type phenotype at a rate of 2.9% in Arabidopsis (Jacobsen and Meyerowitz, 1997). The Cnr mutant is very stable with only 0.1% reversion rate in tomato (Manning et al., 2006). The Epi-df causes a dwarf stature and various floral defects in rice (Oryza sativa) . The reversion of Epi-df occurred in the df progeny at frequencies of 1.97, 0.31, and 0.20%, respectively, in different growing seasons . In our study, the epiallele E epi3 partially reverted with a frequency of 1.0%. The partial revertant Y1130-16 contains only one methylated cytosine and produced seeds with an average of 12.3% erucic acid, suggesting that this partial revertant is a new heritable epimutant. This indicates that DNA methylation can generate multiple epialleles with various expression levels, thereby leading to continuous quantitative variation of a trait. Epigenetic modifications may have played a significant role in quantitative modulation of trait variation.
The epigenetic modification observed here and the mutations induced by the TEs Sal-PIF and Sal-T1 revealed in this study had deleterious effects on the function of FAE1 in yellow mustard. However, such mutants have been maintained in the populations during the speciation and evolution of yellow mustard. Likely, erucic acid content is a neutral trait for plant viability and is not subject to elimination by natural selection. Furthermore, yellow mustard is an obligate outcrossing species in which high levels of heterozygosity could help maintain not only neutral but even slightly deleterious mutations in the species.
To conclude, we have shown that TE insertion and epigenetic modification cause the multiallelic variation in FAE1 in yellow mustard. Our results have demonstrated a comprehensive molecular framework for the interplay of TE insertion, SNP/indel mutation, and epigenetic variation accounting for the broad range of naturally occurring genetic variation in plants.

DNA Extraction
Genomic DNA was extracted from young expanding leaves of the four parental lines Y517, Y496, Y1130, and Y514, F1 and F2 plants using a modified sodium dodecyl sulfate method (Somers et al., 1998).
Cloning the Coding Regions of the FAE1 Alleles E 1 , E 2 , E 3 , and e Primers were designed using the software Primer3 (http://redb.croplab. org/modules/redbtools/primer3.php). Primer pair No 1 (Supplemental Table 1 and Supplemental Figure 2) was designed to clone the coding regions of E 1 , E 2 , E 3 , and e based on the FAE1 sequences in Brassica and the putative FAE1 sequence of yellow mustard. Primer pair No 2 (Supplemental Table 1 and Supplemental Figure 2C) was designed based on the sequence of the 39 coding region of allele E 1 to isolate the partial coding region of allele e. Primer pair No 3 (Supplemental Table 1 and Supplemental Figure 2C) was designed based on the sequence of the partial coding region of e and used to clone the remaining coding sequence of e. Genomic DNA of each of the four parental lines was used as template for PCR amplification with fideliTaq DNA polymerase (Affymetrix) in a thermocycler with 30 cycles of the following program: 94°C for 30 s, 56°C for 30 s, and 68°C for 2 min.
Cloning the 39 and 59 Flanking Sequences of the Coding Regions of the FAE1 Alleles E 1 , E 2 , E 3 , and e The primer pairs used to clone the 39 and 59 flanking sequences of the coding regions of the FAE1 alleles E 1 , E 2 , E 3 , and e are presented in Supplemental Table 1 and Supplemental Figure 2. Primer pair No 4 was designed based on the sequence of the 39 coding region and was used to clone the 39 downstream sequence of each of the four alleles. Primer pair No 5, designed based on the 59 coding region, was used to clone the 59 upstream sequences of E 1 , E 2 , and E 3 . Primer pairs No 6 to 9 (Supplemental Figure 2B) were designed based on the 59 upstream sequence obtained from each previous round of PCR walking to clone the further upstream sequences of E 2 and E 3 . Primer pair No 10, designed based on the promoter sequence of allele E 1 , was used to clone the promoter region of E 2 , E 3 , and e. Primer pair No 11 was designed to clone the DNA fragment between the promoter and 59 flanking sequence of E 2 and E 3 (Supplemental Figure 2B). Primer pair No 12 was designed based on the promoter sequences of E 2 and E 3 and was used to clone upstream sequences of these promoters. Primer pair No 13 was used to clone the DNA fragment between the promoter and 59 flanking sequence of e (Supplemental Figure 2C). PCR walking was performed to clone the 59 and 39 flanking sequences of the coding region of E 1 , E 2 , E 3 , and e according to the protocol of Siebert et al. (1995). The standard protocol from the Clontech kit (Protocol PT 3042, version PR 03300) by Gwyneth Ingram and Karine Coenen was followed to facilitate the PCR walking. To validate the entire sequence including the coding region, 59 and 39 flanking sequences of each of the four alleles E 1 , E 2 , E 3 , and e obtained by PCR walking, primer pair No 14 (WF/WR) (Supplemental Table 1; Figure 1A) was designed to amplify each of the four alleles in their entirety. Genomic DNA of each of the four lines Y517, Y496, Y1130, and Y514 were used as templates with Phusion high-fidelity polymerase (NEB) with 30 cycles of the following program: 98°C for 10 s, 57°C for 30 s, and 72°C for 4 min.

Transformation of Yeast and Fatty Acid Analysis
The coding sequences of E 1 and E 2 were amplified using primer pair No 15 (Supplemental Table 1) and cloned into the pYES2.1/V5-His-TOPO expression vector (Invitrogen), then sequenced to confirm the correct orientation of genes. To determine whether the cloned FAE1 alleles E 1 and E 2 from lines Y517 and Y496 could function as condensing enzymes and were capable of elongation of C18:1 to C20:1 and C22:1, the two constructs pYES2.1/V5-His-TOPO-E 1 and pYES2.1/V5-His-TOPO-E 2 were transformed into Saccharomyces cerevisiae strain Inv Sc1 (Invitrogen) by the lithium acetate method (Ausubel et al., 1995). Yeast cells transformed with the empty vector pYES2.1/V5-His-TOPO plasmid were used as a control. Transgenic cells were screened in complete minimal dropout uracil medium containing 2% raffinose as a carbon source at 30°C. The positive clones were grown at 30°C overnight in minimal media supplemented with 2% raffinose and lacking uracil. The transgenic cell culture was centrifuged, followed by washing, and then used to inoculate 20 mL induction media (minimal media lacking uracil and supplemented with 2% galactose and 1% raffinose) to an OD 600 of 0.5. Cultures were grown overnight. For fatty acid analysis, the resulting cell cultures were pelleted by centrifugation, washed once with the induction media and once with water, and then freeze dried. The dried pellet was transferred to a glass insert. After adding 100 mL hexane and 100 mL of 4% sodium methoxide in methanol, the samples were grinded with a glass rod and then incubated at room temperature for 15 min. Then, 15 mL of 0.2 M sodium phosphate buffer was added to the solution followed by blowing with air stream for 2 min. About 200 mL of heptane was added to the insert and the resulting fatty acid methyl esters were analyzed for fatty acid composition on a Hewlett Packard 6890 gas chromatograph.

Transformation of Arabidopsis and Histochemical GUS Assays
The 59 flanking fragments from the translation start codon for E 1 and E 2 were cloned using primer pairs No 16 and 17 (Supplemental Table 1) and were inserted into the plant expression vector pBI101 upstream of the GUS gene. The resulting constructs pBI101-Pro E1 -GUS with the 59 flanking fragment of E 1 and pBI101-Pro E2 -GUS with the 59 flanking fragment of E 2 were transformed into electrocompetent Agrobacterium tumefaciens cells GV3101 by electroporation according to the manufacturer's instructions. Transformation of wild-type Arabidopsis thaliana plants was performed using the Agrobacterium-mediated transformation method described by Clough and Bent (1998). Transgenic plants were screened and analyzed according to Jako et al. (2001). Embryos of 9-dold, leaf, stem, bud, and flower tissues of the transgenic plants were stained overnight in solution containing 0.5 mg/mL 5-bromo-4-chloro-3indolyl-b-D-glucuronide, 50 mM Na phosphate buffer, pH 7.0, 1 mM K 3 Fe (CN) 6 , 1 mM K 4 Fe(CN) 6 , and 20 mM EDTA. The staining of the embryo and vegetative tissues was observed and photographed using a Zeiss stereomicroscope with a color CCD camera.

RT-PCR and Quantitative Real-Time PCR
Total RNA was extracted from leaf, stem, flower bud, flower, and embryos of Y517, Y496, Y1130, and Y514 using an RNeasy plant mini kit (Qiagen). Total RNA was extracted from only 21-d-old embryos of the 5-azaCinduced and spontaneous revertants of Y1130. Extracted RNA was treated with DNase I (Ambion), and cDNA was synthesized using the ReverTra Ace-a-First-Strand cDNA synthesis kit (Fermentas) according to the manufacturer's instruction. Primer pair No 18 (Supplemental Table 1) was designed based on the coding sequences (position 1 to 928) of the FAE1 alleles E 1 , E 2 , and E 3 . The expression of the FAE1 alleles was then analyzed by RT-PCR using primer pair No 18 with 35 amplification cycles.
To detect differences in transcription levels among FAE1 alleles E 1 , E 2 , and E 3 , real-time qPCR analysis was conducted on embryos harvested at 14, 21, 28, and 35 d after pollination from plants of Y517, Y496, and Y1130. The 21-d-old embryos of 5-azaC-induced revertants and spontaneous revertant of Y1130 were also used for qPCR analysis to determine the expression level of FAE1 allele E 3 . Primer pair No 19 (Supplemental Table 1) specific for FAE1 was used to amplify ;100-bp fragments (nucleotides 865 to 964) from the cDNA of 14-to 35-d-old embryos in Y517, Y496, and Y1130, the cDNA of 21-d-old embryos of the 5-azaC-induced revertants and the spontaneous revertant of Y1130. The qPCR analysis was performed with SsoFast EvaGreen supermix (Bio-Rad) according to the manufacturer's instructions using a Bio-Rad CFX96 system. Primer pair No 20 (Supplemental Table 1) specific for Actin2 (FG576123) was used as an internal control for normalization. Three separate first-strand cDNA reactions were analyzed in duplicate for each sample, and expression levels were calculated as described by Livak and Schmittgen (2001).
Sequence Analysis of the Inserted DNA Fragments in the 59 UTR of FAE1 Alleles E 2 and E 3 and in the Coding Region of FAE1 Allele e The open reading frames and deduced amino acid sequences of the inserted fragments in the 59 UTR of FAE1 alleles E 2 and E 3 and in the coding region of FAE1 allele e were predicted using the GenScan Web server at MIT (http://genes.mit.edu/GENSCAN.html). The conserved domains of the deduced amino acid sequences were identified using RPS-BLAST (reverse position-specific BLAST) (http://www.ncbi.nlm.nih. gov/Structure/cdd/docs/cdd_search.html). The PBS of the inserted fragment in the 59 UTR of E 2 and E 3 was identified by aligning the sequence of the 59 LTR to the 39 region of tRNAs from a tRNA database (http://lowelab.ucsc.edu/GtRNAdb/). The PPT was determined by examining the purine-rich sequence in the upstream region of the 39 LTR. The amino acid sequence alignment of maize PIF and the inserted fragment in the coding region of FAE1 allele e was performed by the ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/).

Bisulfite Sequencing
Accurate analysis of DNA methylation by bisulfite sequencing depends on the complete conversion of all cytosines into uracil. Wang et al. (2011) developed a PCR-based assay for the mitochondrial gene ATP1 (ATPase SUBUNIT 1) and demonstrated its efficacy as a bisulfite conversion control in Brassica rapa and other plant taxa. The two primer pairs for the ATP1 provide a reliable control across a representative range of dicot and monocot angiosperm species. Therefore, the ATP1 of yellow mustard was used as a negative control to check if the bisulfite chemical reaction was complete. The ATP1 (227 bp) was amplified using primer pair No 21 (Supplemental Table 1; Wang et al., 2011). In the wild-type Arabidopsis Columbia-0, AtSN1 (A. thaliana SINE like element 1), a putative transposable retroelement located on chromosome 3, contains 18 methylated cytosines in the internal region between the long terminal repeats (Kuhlmann and Mette, 2012). AtSN1, with 18 known methylated cytosines, was used as a positive control in our bisulfite sequencing experiment. The AtSN1 (nucleotides 15,805,617 to 15,805,773, NCBI nucleotide sequence NC003074) was amplified using primer pair No 33 (Supplemental Table 1). Bisulfite sequence alignment of the AtSN1 (positive control) is shown in Supplemental Figure 9.
Bisulfite sequencing was performed on genomic DNA from three plants of both Y517 (E 1 E 1 ) and Y496 (E 2 E 2 ), four plants of Y1130 (E 3 E 3 ), 5-azaC-induced revertants, and spontaneous revertant of Y1130. Genomic DNA from the embryos of each plant was bisulfite treated using the EpiTect Bisulfite kit (Qiagen). The candidate methylated coding and promoter regions of the FAE1 alleles were amplified using bisulfite primers No 22-27 and No 28-30, respectively, designed using MethPrimer software, and the resulting fragments were cloned into the pGEM-T vector (Promega). Primer pairs No 22-27 (Supplemental Table 1) were used to amplify the coding region (nucleotides 1 to 1515) of E 1 , E 2 , and E 3 . Primer pair No 28 (Supplemental Table 1) was designed to amplify the promoter sequence and partial coding region (nucleotides 2143 to 56) of E 1 . Primer pair No 29 (Supplemental Table 1) was designed to amplify the promoter sequence (nucleotides 25042 to 24900) and retrotransposable element Sal-T1 (nucleotide 24899 to 24843) of E 2 and E 3 . Primer pair No 30 (Supplemental Table 1) was designed based on the promoter sequence of E 1 and was used to detect the methylation status of the promoter region (nucleotides 2335 to 279) of E 1 and the promoter regions (nucleotides 25234 to 24978) of E 2 and E 3 . At least 10 independent clones of each fragment generated by each primer pair were sequenced to determine the number and position of methylated residues for each plant. Sequence analysis was performed according to the Web-based tool Kismeth (Gruntman et al., 2008).

5-Azacytidine Treatment
Forty-five seeds of the low erucic acid line Y1130 were treated with 1 mM 5-azaC (Sigma-Aldrich), a demethylation agent. A stock solution of 15 mM 5-azaC was prepared in 50 mM MES buffer (PH 6.3). The seeds were placed evenly on a filter paper in a Petri dish to which 4.5 mL 1 mM 5-azaC solution was added. The Petri dish was then incubated in the dark at room temperature for 3 d. The 5-azaC-treated seeds were transferred to 50-mL tubes, rinsed with water for eight times, and then transferred to soil for germination and growth in the greenhouse. The resulting 5-azaC-treated plants were bagged and bud-pollinated to produce self-pollinated seeds, since yellow mustard is an obligatory out-crossing species with a self-incompatible reproduction system. The self-pollinated 21-d-old embryos of each plant were sampled for DNA methylation and real-time qPCR analysis. The self-pollinated seeds of each plant were used for fatty acid analysis.

Cosegregation of the FAE1 Allele-Specific Markers with Erucic Acid Levels
Primer pair No 31 (Supplemental Table 1) was designed based on the sequences of the promoter and the transposon Sal-PIF inserted in the coding region of FAE1 allele e. Primer pair No 32 (Supplemental Table 1) was designed based on the promoter sequence and the sequence of the retrotransposon Sal-T1 in the 59 UTR of FAE1 alleles E 2 and E 3 . The cosegregation of the marker specific for the allele e and zero erucic content was studied in 96 F2 plants derived from the cross Y514 (ee) 3 Y517 (E 1 E 1 ), while the cosegregation of the marker specific for the allele E 2 with medium erucic content was studied in 70 F2 plants of the cross Y496 (E 2 E 2 ) 3 Y517 (E 1 E 1 ). Each PCR (20 µL) contained 13 PCR standard buffer (NEB), 200 µM each deoxynucleotide triphosphate, 0.1 µM of each forward and reverse primer, 1 unit of Taq polymerase (NEB), and 50 ng of genomic DNA. PCR was performed with an initial denaturation at 94°C for 3 min followed by 30 cycles of 30 s at 94°C, 30 s at 58°C, and 1 min at 72°C, followed by a final extension cycle of 72°C for 7 min.

Fatty Acid Analysis
The seed erucic acid contents of lines Y517, Y496, Y1130, the 5-aza-Cinduced revertants, and spontaneous revertant of Y1130 were determined using the half-seed technique (Downey and Harvey, 1963) and the gas chromatographic method of Thies (1971).

Accession Numbers
The GenBank accession numbers for the nucleotide sequences of the FAE1 allele E 1 , E 2 (E 3 ), and e in yellow mustard are KJ817378, KJ817379, and KJ817380, respectively. The nucleotide sequence of Sal-PIF, a 3100bp PIF/Harbinger-like DNA transposon, is KJ817381. The nucleotide sequence of Sal-T1, a 4863-bp Copia-like retrotransposon, is KJ817382. The GenBank accession number of the putative FAE1 sequence of yellow mustard is AY888040. The GenBank accession number of the FAE1 promoter region from the high erucic acid B. napus line Askari is AF275254. The GenBank accession number of the putative retroelement pol polyprotein is AAC67205. The GenBank accession numbers of the FAE1 genes in B. juncea, B. napus, B. rapa, B. oleracea, and Arabidopsis are AJ558198, AF490459, AF49041, AF490460, and Q38860, respectively. The GenBank accession number of the amino acid sequence maize PIF is AF412282.

Supplemental Data
The following materials are available in the online version of this article. Supplemental Figure 8. PCR Amplification of the Entire FAE1 Alleles E 1 , E 2 , E 3 , and e Using the Primer Pair WF/WR No 14 (Supplemental Table 1).
Supplemental Figure 9. DNA Methylation in the Internal Region between the Long Terminal Repeats of AtSN1 (Positive Control).
Supplemental Figure 10. DNA Methylation in the Coding Regions of the FAE1 Alleles E 1 , E 2 , and E 3 .
Supplemental Figure 11. Bisulfite Sequence Alignment of ATP1 (Control Gene) and the FAE1 Allele E 1 in Line Y517.
Supplemental Figure 12. Bisulfite Sequence Alignment of ATP1 (Control Gene) and the FAE1 Allele E 2 in Line Y496.
Supplemental Figure 13. Bisulfite Sequence Alignment of ATP1 (Control Gene) and the FAE1 Allele E 3 in Line Y1130.
Supplemental Figure 14. Bisulfite Sequence Alignment of ATP1 (Control Gene) and the FAE1 Allele E 3 in the 5-azaC Induced Revertants of Y1130.
Supplemental Figure 15. Bisulfite Sequence Alignment of ATP1 (Control Gene) and the FAE1 Allele E 3 in the Spontaneous Revertant Y1130-16.
Supplemental Table 1. Primer Pairs Used in This Study.