Tandemly repeated endogenous genes are common in plants, but their transcriptional regulation is not well characterized. In maize, the P1-wr allele of pericarp color1 is composed of multiple copies arranged in a head-to-tail fashion. P1-wr confers a white kernel pericarp and red cob glume pigment phenotype that is stably inherited over generations. To understand the molecular mechanisms that regulate tissue-specific expression of P1-wr, we have characterized P1-wr*, a spontaneous loss-of-function epimutation that shows a white kernel pericarp and white cob glume phenotype. As compared to its progenitor P1-wr, the P1-wr* is hypermethylated in exon 1 and intron 2 regions. In the presence of the epigenetic modifier Ufo1 (Unstable factor for orange1), P1-wr* plants exhibit a range of cob glume pigmentation whereas pericarps remain colorless. In these plants, the level of cob pigmentation directly correlates with the degree of DNA demethylation in the intron 2 region of p1. Further, genomic bisulfite sequencing indicates that a 168-bp region of intron 2 is significantly hypomethylated in both CG and CNG context in P1-wr* Ufo1 plants. Interestingly, P1-wr* Ufo1 plants did not show any methylation change in a distal enhancer region that has previously been implicated in Ufo1-induced gain of pericarp pigmentation of the P1-wr allele. These results suggest that distinct regulatory sequences in the P1-wr promoter and intron 2 regions can undergo independent epigenetic modifications to generate tissue-specific expression patterns.
THE variants of genetic loci generated through epigenetic modifications, or epimutations, are known as epialleles. Epialleles may be metastable and revert to wild type or produce novel phenotypes. Epimutations have been identified in several important plant biological processes including floral symmetry and development (Jacobsen and Meyerowitz 1997; Cubas et al. 1999), disease resistance (Stokes et al. 2002), and plant and seed pigmentation (Das and Messing 1994; Dorweiler et al. 2000; Hoekenga et al. 2000). Stable epialleles in natural populations provide new sources of variation as well as tools to study mechanisms that generate these epialleles (Kishimoto et al. 2001; reviewed in Kalisz and Purugganan 2004).
Formation of epialleles has been attributed to two mutually dependent epigenetic processes: changes in DNA methylation and chromatin modifications (Kalisz and Purugganan 2004; reviewed in Richards 2006). DNA methylation, covalent attachment of methyl group to cytosine base, is a heritable genetic modification in eukaryotes. A growing body of evidence has implicated DNA methylation as a key epigenetic regulator of stable gene expression patterns (see reviews by Chan et al. 2005; Richards 2006). In many cases, regulation of tissue-specific gene expression has been attributed to developmental changes in cytosine methylation (Bao et al. 2004; Kinoshita et al. 2004). In plants, the role of DNA methylation has also been established in transcriptional and post-transcriptional gene silencing (Vaucheret and Fagard 2001; Vaucheret et al. 2001), inactivation of transposons and transgenes (Chandler and Walbot 1986; Banks et al. 1988; Ingelbrecht et al. 1994), genomic imprinting (Alleman and Doctor 2000; Gehring et al. 2006), and paramutation (Walker 1998; Sidorenko and Peterson 2001).
In maize, anthocyanin and phlobaphene pigments have been used as convenient markers to identify spontaneous epimutations in loci that encode transcription factors; meiotically stable epialleles have been isolated and characterized for booster1 (b1), red1 (r1), purple plant 1 (pl1), and pericarp color1 (p1) (Brink 1956; Cocciolone and Cone 1993; Das and Messing 1994; Stam et al. 2002). The p1 gene encodes a Myb transcription factor that regulates biosynthesis of brick red pigments known as phlobaphenes. These p1-regulated 3-deoxyflavonoid pigments can be observed in kernel pericarp, cob and tassel glumes, husk, and leaf sheath (Coe et al. 1988; Styles and Ceska 1989). Alleles of p1 are identified by a two-letter suffix that denotes their expression in pericarp and cob glume (Emerson 1917; Anderson 1924; Brink and Styles 1966). For example, P1-rr conditions red pericarp and red cob glume phenotype while P1-wr specifies white (colorless) pericarp and red cob glume pattern. The mechanisms of tissue-specific variation observed in P1-rr and P1-wr are not completely understood. These alleles differ considerably with respect to their gene structure: P1-rr is a single-copy allele in which the coding sequence is flanked by two 5.2-kb direct repeats (Lechelt et al. 1989; Grotewold et al. 1991), whereas P1-wr contains six or more copies of 12.6 kb each that are tandemly arranged in a head-to-tail fashion (Chopra et al. 1998). However, these alleles share high sequence similarities in their upstream promoter (99%) and coding (99.9%) regions (Chopra et al. 1998). Unique expression patterns of these alleles have been proposed to be regulated through epigenetic mechanisms: the multicopy P1-wr is hypermethylated as compared to the single-copy P1-rr (Chopra et al. 1998). The hypermethylated state and the white pericarp/red cob glume phenotype of P1-wr are stably inherited. However, in the presence of a trans-acting modifier Unstable factor for orange1 (Ufo1), P1-wr becomes less methylated and shows increased p1 transcription (Chopra et al. 2003). As a result, P1-wr Ufo1 plants exhibit gain of pigmentation in kernel pericarp, cob and tassel glumes, husk, and leaf sheath. The Ufo1-induced phenotypes show a range of pigmentation that positively correlates with the degree of demethylation of the P1-wr repeat complex. Although the effect of Ufo1 is not fully penetrant, it provides an effective genetic and molecular tool to characterize mechanisms that regulate tissue-specific expression patterns of p1.
Here, we describe a spontaneous epiallele of P1-wr, designated as P1-wr*, that specifies a white pericarp and white cob glume phenotype. Our results show that the loss of cob glume pigmentation phenotype is associated with hypermethylation of cytosine residues in the exon 1 and intron 2 regions of P1-wr*. In addition, Ufo1 interacts with the silent P1-wr* in a tissue-preferred manner that results in gain of cob (but not pericarp) pigmentation. Furthermore, the Ufo1-induced gain of cob function is correlated with demethylation of sequences in intron 2 to a level similar to that of standard P1-wr. Our study suggests that the intron 2 region may contain cis-regulatory elements and that epigenetic modification of these sequences can impact the cob-specific expression of P1-wr.
MATERIALS AND METHODS
Recombinant inbred (RI) line RI41 and its sibling lines were obtained from Benjamin Burr, Brookhaven National Laboratories, New York. These RI lines belong to a mapping population (T × CM family) developed from a cross of T232 × CM37 (Burr et al. 1988). The T232 inbred line carries a functional P1-wr allele and has a white (colorless) kernel pericarp and red cob glume (WR) phenotype, while the CM37 inbred has a p1-ww allele and exhibits white (colorless) kernel pericarp and white cob glume (WW) phenotype (Figure 1). Genetic and molecular analyses (see results) indicate that the RI41 line contains an epigenetically silenced P1-wr allele referred to here as P1-wr*. The RI41 line and its selfed progeny plants maintain a WW phenotype, while a low percentage of ears (∼2.5%) show pink or red patches of cob glumes. Biochemical analysis using acid–methanol extracts of these tissues indicated the presence of low levels of flavan-4-ols (data not shown) that are precursors of phlobaphenes (Styles and Ceska 1989). Plants grown from kernels attached to pink/red cob glume areas did not produce ears with pigmented cobs, indicating that these pigmented areas are not associated with heritable reversions. Further, genetic tests (data not shown) established the following: (1) the structural genes c2 and a1 that encode flavonoid biosynthetic enzymes are in a functional condition in RI41; (2) the F1 plants produced by crossing RI41 with P1-wr in three inbred genetic backgrounds (T232, W22, and W23) have colorless pericarp and red cobs; thus, the WW phenotype of RI41 is recessive to P1-wr; and (3) the progeny plants produced by test crossing F1 plants of genotype P1-wr*/P1-wr with a standard p1-ww allele segregate WR and WW phenotypes in a 1:1 ratio.
A stock containing Ufo1 in a P1-wr background was obtained from Derek Styles, University of Victoria (Victoria, BC, Canada) and has previously been described (Chopra et al. 2003). The Ufo1 factor was introgressed into an inbred line 4Co63 (genotype p1-ww c1 r-r; National Seed Storage Laboratory, Fort Collins, CO) and a p1-ww Ufo1 homozygous stock was developed by self fertilization for several generations. Since Ufo1 does not induce any pigmentation change in p1-ww plants, the presence of Ufo1 was tested by outcrossing to P1-wr plants. Individual plants from p1-ww Ufo1 stock were crossed to W22 or W23 (genotype P1-wr; Maize Genetics Cooperation Stock Center, Urbana, IL) and 20 plants were grown from each outcross. Orange pigmentation of ligules, leaf sheath, husk, and kernel pericarp in these progeny plants confirmed the presence of Ufo1 in p1-ww Ufo1 stock.
Homozygous p1-ww Ufo1 plants were crossed to RI41 and a total of 133 F1 plants derived from seven crosses were grown. These F1 plants were crossed with p1-ww[4Co63] to develop test-cross progenies. Twenty-four test-cross ears showing a gain of cob glume pigmentation phenotypes (red or pink) were further selected to obtain progeny plants for molecular characterization and inheritance studies. From each test-cross progeny, six to eight plants were used for DNA methylation and genotypic analysis. Plants that did not survive until harvest were excluded from the analysis. Phenotypic data from 117 progeny plants were obtained and correlated with the results of genotypic and DNA methylation analyses to determine the heritability of the Ufo1 effect on P1-wr*.
DNA gel blot hybridizations:
Plant genomic DNA was isolated using a modified CTAB method (Saghai-Maroof et al. 1984). Restriction enzyme digestions were performed to completion using enzymes, reagents, and incubation conditions from Promega (Madison, WI). Restricted genomic DNA was fractionated on agarose gels, transferred to nylon membranes, and the resulting blots were sequentially probed with DNA fragments representing different regions of p1. DNA gel blot hybridizations were performed using a hybridization mixture containing dextran sulfate (10%), NaCl (1 m), SDS (1%), Tris–HCl (10 mm), and 0.25 mg/ml salmon sperm DNA. Blots were prehybridized at 65° for 6 hr in the hybridization buffer followed by hybridization in the same buffer containing a 32P-labeled probe for ∼20 hr at 65°. Blots were washed twice in 0.1× SSC and 0.5% SDS at 65° for 15–30 min and exposed to X-OMAT film (Kodak, Rochester, NY). Before rehybridization, the blots were stripped by washing for 15 min in a boiling solution of 0.1% SDS.
Genomic bisulfite sequencing:
DNA methylation of target regions was assayed by genomic bisulfite sequencing. These included the top DNA strand from a 333-bp region from the promoter (region I), a 312-bp region near the 5′ end of intron 2 (region II), and a 435-bp region near the 3′ end of intron 2 (regions III and IV). For bisulfite treatment, 8 μg of leaf genomic DNA from the appropriate genotypes were completely digested with restriction enzymes that cut outside the regions of interest. Sodium bisulfite treatment was performed on the basis of a previously published protocol (Jacobsen et al. 2000) with minor modifications. Primers specifically designed to amplify DNA modified with sodium bisulfite were used for PCR. Sequences of primers used for bisulfite genomic sequencing are available upon request. The PCR products were gel purified and cloned using a TOPO TA cloning kit (Invitrogen, Carlsbad, CA), and ∼25 individual clones of each ligation reaction were subjected to DNA sequencing.
RI41 carries an epiallele of P1-wr:
The presence of occasional areas of red phlobaphene pigments on cobs of RI41 plants prompted us to investigate the gene structure of the p1 allele in this line. Gel blots prepared from leaf DNA of RI41, T232, and W23 digested with several restriction enzymes were hybridized with three p1 probe fragments (Figure 2). The restriction patterns produced by SacI and KpnI digests were similar in T232, RI41, and W23. For example, probe fragment 15, corresponding to the p1 distal enhancer region, hybridized to a SacI band (see lanes labeled “Sc” in Figure 2A) of 0.87 kb in all three genotypes. Hybridization of the same blot with another fragment from the 5′ upstream promoter region (fragment 6) gave two prominent bands with expected sizes of 4.08 and 0.87 kb in all three genotypes. Similarly, SacI digestion produced a common 5.98-kb band upon hybridization with probe fragment 8B that corresponds to the p1 intron 2 region. In summary, no restriction fragment length polymorphism (RFLP) differences were detected for p1 in comparisons of RI41 with standard P1-wr present in T232 or W23 inbred lines. Furthermore, no differences in p1-hybridizing band intensities were observed, indicating that copy number of p1 tandem repeats was not affected, for example, by recombination between closely linked repeated copies. Taken together, these DNA gel blot analyses indicate that RI41 carries a P1-wr allele that is structurally similar to standard P1-wr. To distinguish it from the standard P1-wr allele present in W23 or T232, the RI41 allele is designated as P1-wr*.
In the absence of any obvious RFLP, we tested the possibility of epigenetic silencing of P1-wr*. DNA methylation of P1-wr and P1-wr* was compared using methylation-sensitive enzymes SalI and HpaII (Figure 2). P1-wr[W23] was used as a reference since DNA methylation of this allele has been extensively studied (Chopra et al. 1998, 2003). As expected, hybridization of each of the three probes gave an ∼12.6-kb SalI band in P1-wr (Figure 2A). This band results from cleavage of two SalI sites in intron 2 of each of the adjoining P1-wr gene copies; the other two internal SalI sites are hypermethylated in all copies and are shown as SalI* in Figure 2B (Chopra et al. 1998). In P1-wr*, however, a higher molecular weight band (>12.6 kb) was observed upon hybridization with each probe, indicating that additional SalI sites are methylated. HpaII digests of P1-wr in W23 or T232 gave two bands with probes 15 and 6: one prominent band of ∼7.9 kb and a second relatively weak band of ∼12.0 kb (Figure 2A). The ∼7.9-kb band is produced by digestion at HpaII site(s) in two distinct regions of the p1 gene (cluster 3 and cluster 7; see Figure 2B). The ∼12.0-kb band is derived by cutting of HpaII sites in the same region (cluster 7) of two adjoining copies, indicating that HpaII sites in the intervening area (cluster 3) are partially methylated in some of the gene copies and/or cells. Compared with P1-wr, hybridization of P1-wr* with probes 15 and 6 produced an ∼12-kb band that was more prominent than the ∼7.9-kb band. This pattern could result from increased methylation of HpaII sites in cluster 3 in P1-wr*. This result was confirmed with intron 2 probe fragment 8B, which hybridized to ∼12.0- and ∼4.5-kb bands in P1-wr but gave only the ∼12.0-kb band in P1-wr*. Thus, the absence of the ∼4.5-kb band in P1-wr* further established that the HpaII sites in cluster 3 are more methylated than those in P1-wr. In summary, these results indicated that P1-wr* is more methylated than P1-wr. However, it should be noted that the large number of HpaII sites (57 per P1-wr gene copy) together with the multiple methylation patterns hampers the precise determination of methylation status by DNA gel blots.
Ufo1 reverts the P1-wr* epiallele to P1-wr:
Previous results showed that Ufo1 interacts with wild-type P1-wr to induce pigmentation of kernel pericarp and vegetative tissues (Chopra et al. 2003). To test the interaction of P1-wr* with Ufo1, we crossed P1-wr* with p1-ww Ufo1 (see materials and methods; Figure 3). The resulting F1 plants were screened for presence or absence of phlobaphene pigments in the pericarp, cob and tassel glumes, leaf sheath, husk, and silk. Of 133 F1 plants screened, 46 (35%) had white pericarp and pink or red cob pigmentation (expressers), while 87 (65%) had white pericarp and white cobs (nonexpressers). Similarly, the tassel glume margins in the expresser group also showed red or pink pigmentation that correlated with cob color phenotype of that plant. In addition, RNA gel blot results identified p1 transcripts in the cobs of expressers, while no transcripts were detected in nonexpressers (data not shown). To summarize, these results established that the silent state of P1-wr* is reactivated upon crossing with the Ufo1 stock. Surprisingly, unlike the P1-wr × Ufo1 interaction, none of the P1-wr* Ufo1 expresser plants showed any gain of pigmentation in pericarp, leaf sheath, husk, or silk. These results indicate that Ufo1 affects P1-wr* differently than P1-wr; however, the incomplete penetrance of Ufo1 persists in both the alleles.
P1-wr* is hypomethylated in the presence of Ufo1:
We have previously demonstrated that the Ufo1-induced gain of pericarp, cob, and plant body pigmentation is correlated with hypomethylation of P1-wr (Chopra et al. 2003). We tested if gain of cob pigmentation in P1-wr* Ufo1 plants is also associated with hypomethylation of P1-wr*. DNA gel blots prepared from HpaII-digested leaf DNA of the F1 plants were hybridized with p1 genomic probe fragment 15, and the results are presented in Figure 4. On the basis of DNA methylation and cob color, F1 plants could be divided into three groups. The first group (Figure 4A, lanes 6 and 7) was characterized by the presence of a very intense ∼7.9-kb band and a faint ∼12.0-kb band, a pattern similar to that of the wild-type P1-wr (see lane 2). As expected, the cob pigment phenotypes of this category also resembled that of P1-wr (Figure 4B, sections 2 and 6). The second group had moderate methylation levels indicated by roughly equal intensities of ∼7.9- and ∼12.0-kb bands (Figure 4A, lanes 4 and 5). The cobs of these plants showed pink to red variegations or sharp red/white sectors that covered the entire length of the ear (Figure 4B, compare sections 4 and 5). Interestingly, plants belonging to the third group had the highest degree of DNA methylation as shown by the absence of the ∼7.9-kb band and presence of high molecular weight bands (Figure 4A, lanes 8 and 9). These individuals produced ears with white cob phenotypes (Figure 4B, section 9). To rule out any effect of the 4Co63 genetic background per se on P1-wr*, F1 plants of several independent crosses (P1-wr* × p1-ww[4Co63]) were also examined for their DNA methylation levels; no changes in cob phenotype or DNA methylation of P1-wr* were observed (data not shown). In summary, these results established that Ufo1-induced gain of cob color in P1-wr* is accompanied by hypomethylation of the P1-wr sequence and that the extent of demethylation is positively correlated with the level of cob pigmentation.
Gain of cob color correlates with DNA hypomethylation of the intron 2 region:
Our results showed that, compared with P1-wr, P1-wr* is hypermethylated at HpaII site 2 and clusters 3 and 7 (see Figure 2B for position of these sites/clusters). Exposure of P1-wr* to Ufo1 resulted in gain of cob pigmentation that was associated with demethylation of P1-wr* (see Figure 4). To delineate specific sequence(s) of P1-wr* that may be playing a role in Ufo1-induced cob-specific expression, methylation of candidate HpaII site 2 and clusters 3 and 7 was assayed. We used DraI in combination with HpaII and probed these digests with p1 intron 2 probes 8B and 8C (Figure 5). Since DraI endonuclease is methylation insensitive, probe 8B detected an expected band of 2.6 kb in single digests for all three genotypes (Figure 5A). This probe also hybridizes to two other bands of ∼9.0 and ∼3.0 kb that may come from a polymorphic copy (copies) of the P1-wr multicopy complex (S. Chopra and T. Peterson, unpublished results). The hybridization of HpaII-digested P1-wr produced a 4.5-kb band, which was absent from P1-wr* because of hypermethylation of cluster 3 sites. The presence of a 4.2-kb band in P1-wr* Ufo1 indicated hypomethylation of HpaII sites in clusters 3 and 7 (see below). In HpaII and DraI double digests of P1-wr, probe 8B hybridized to two prominent bands of 2.3 and 2.1 kb. The 5′ ends of these bands are produced by digestion at two partially methylated HpaII sites in the middle of cluster 7 and the 3′ ends result from digestion by DraI (see Figure 5, A and C). In P1-wr*, the presence of a prominent 2.6-kb band and a faint 2.3-kb band established that HpaII sites in cluster 7 were highly, but not completely, methylated. In two P1-wr* Ufo1 expressers, the HpaII + DraI digestion generated two fragments of 2.1 and 2.0 kb. The 2.1-kb fragment, which is also observed in P1-wr, indicates hypomethylation of HpaII sites within cluster 7. Presence of the unique 2.0-kb band indicated that HpaII sites in cluster 7 were slightly less methylated in P1-wr* Ufo1 as compared to P1-wr plants. To more precisely determine the methylation status of HpaII site 2 and sites in cluster 3, the blot was reprobed with fragment 8C. DraI produced an expected fragment of 2.8 kb in all the genotypes. Double digests of P1-wr with HpaII and DraI produced two bands of 2.2 and 0.59 kb, indicating hypomethylation of one or both overlapping HpaII sites in cluster 3 (shown as hatched circles in Figure 5C). Additionally, in P1-wr, the presence of faint bands of 2.8 and 0.86 kb indicated that HpaII site 2 and sites in cluster 3 are partially hypomethylated. In P1-wr*, DraI and HpaII + DraI digests produced similar banding patterns, establishing that the HpaII sites (site 2 and cluster 3) are completely methylated. In P1-wr* Ufo1 expressers, DraI + HpaII digests produced 2.2- and 0.59-kb fragments, indicating that one or both sites present in cluster 3 became hypomethylated. Taken together, these results indicated that the gain of cob pigmentation in P1-wr* Ufo1 plants strongly correlates with hypomethylation of HpaII sites in the 3′ end of intron 2.
To map specific cytosine methylation changes in P1-wr* Ufo1 plants, we performed genomic bisulfite sequencing (Figure 6; see also supplemental Table 1 at http://www.genetics.org/supplemental/). Four representative regions spanning the 12.6 kb of a single P1-wr gene repeat were assayed. Region I is part of a distal enhancer element while region II is present at the 5′ end of intron 2; these regions have previously been implicated in gain of pigmentation in P1-wr Ufo1 plants (Chopra et al. 2003). Selection of regions III and IV was based on their HpaII methylation differences between P1-wr* and P1-wr* Ufo1 expressers (see cluster 3 in Figure 5C). The region encompassing cluster 7 was not selected for bisulfite assay because its methylation status did not produce correlative cob glume pigmentation patterns in different genotypes under study. Of the four regions studied, region III (168-bp sequence) had dramatic differences in cytosine methylation among the three genotypes (Figure 6B). This region had very high CG and moderately high CNG methylation in P1-wr* as compared to P1-wr in which methylation was negligible. Presence of Ufo1 resulted in complete loss of CG and CNG methylation in region III of P1-wr*. Further, in P1-wr and P1-wr* Ufo1 plants, regions III and IV were separated by a sharp boundary of DNA methylation; region IV had exceptionally elevated CG and CNG methylation as compared to region III. This observation indicated that methylation of region IV does not affect cob glume pigmentation. Interestingly, region IV was the only region in the gene that has substantial asymmetric (CHH) methylation in P1-wr and P1-wr* Ufo1, whereas this region in P1-wr* was not methylated in CHH context. The bisulfite assay showed that all three genotypes had high and nearly similar levels of CG and CNG methylation in regions I and II. In summary, gain of cob pigmentation in P1-wr* Ufo1 plants was accompanied by a sharp decrease in CG and CNG methylation of the 168-bp sequence (region III) of intron 2 that was highly methylated in P1-wr*. These results demonstrated that Ufo1 restores methylation of P1-wr* to a state similar to that of wild-type P1-wr.
Ufo1 is required for establishment but may not be needed for maintenance of cob expression:
To test if Ufo1-induced changes in cob glume color and DNA methylation of P1-wr* Ufo1 expressers were heritable, F1 plants were test crossed with p1-ww[4Co63] (see materials and methods). Phenotypic analysis of 117 test-cross progeny plants revealed three classes of cob color phenotypes: white, light to dark pink, and red (see Table 1). To genotype these plants for the p1 allele and to compare their DNA methylation, gel blots prepared from HpaII-digested leaf genomic DNA were hybridized with p1 probe fragment 6, and a representative blot is shown in Figure 7. Since p1-ww/p1-ww plants do not show P1-wr-specific HpaII bands (∼7.9 and ∼12.0 kb), we could easily genotype P1-wr*/p1-ww and p1-ww/p1-ww plants. This allowed separation of test-cross progenies with white cob phenotype into two classes: plants carrying p1-ww (Figure 7, lanes A) and those carrying silenced P1-wr* (Figure 7, lanes W). Among expresser plants, the intensity of the ∼7.9- and ∼12.0-kb HpaII bands established that P1-wr* in red cob plants (Figure 7, lanes R) was hypomethylated as compared to that in pink cob plants (Figure 7, lanes P). These combined phenotypic and molecular data allowed us to perform segregation analysis (Table 1). The progeny plants segregated in a 1:1 (P1-wr*/p1-ww:p1-ww/p1-ww) ratio as expected from a test cross (64:53; χ2 = 1.03; P < 0.05). Among P1-wr*/p1-ww plants, half of the individuals were expected to carry Ufo1 and show gain of cob pigmentation assuming that Ufo1 is fully penetrant. However, within this class, significantly more plants showed red or pink cob pigmentation than expected [40/64 (62%); χ2 = 4.0; P < 0.05]. This result indicated that, in some of these plants, the P1-wr* allele retained its active state even though Ufo1 segregated away. In other words, some of the P1-wr*/p1-ww; ufo1/ufo1 plants had pigmented cobs, indicating that P1-wr* was functional in these plants. This observation indicated that, following establishment of the active state of P1-wr*, Ufo1 may not be required for maintenance of cob pigmentation. However, further testing is needed to establish the absence of Ufo1 from these segregants.
To elucidate the epigenetic mechanisms that establish and maintain stable allele-specific patterns of expression, we have characterized P1-wr*, a natural loss-of-function epiallele of P1-wr. P1-wr* shows a low frequency of mitotic instability in the form of red/white somatic sectors on cob glumes; however, it is meiotically stable as indicated by the absence of germinal red cob revertants. Our results established that P1-wr* exhibits a greater degree of DNA methylation than the standard P1-wr allele. The genesis of the P1-wr* epiallele seems to be a rare spontaneous event, because sibling recombinant inbred lines carrying P1-wr (WR phenotype) did not show any detectable methylation or phenotypic alterations (data not shown). In plants, several environmental, physiological, and genetic modifications may cause epigenetic instability. For example, cold temperature can alter DNA methylation and chromatin structure, thereby generating natural epialleles (Steward et al. 2002; Bastow et al. 2004). Additionally, chemical mutagens (Axtell and Brink 1967) as well as regeneration of plants through tissue culture have been reported to form new epialleles with DNA methylation changes (Kaeppler and Phillips 1993; Kaeppler et al. 2000). Maize is an open-pollinated crop and it is possible that genetic stress as a result of the inbreeding during development of the recombinant inbred lines led to the epigenetic modifications of P1-wr. Indeed, both inbreeding depression and heterosis have been proposed to involve an epigenetic component; self pollination may lead to gene silencing while outcrossing may relieve silencing (Auger et al. 2004). Such a phenomenon was observed at the maize regulatory locus pl1 in which an epiallele Pl′-mah had increased expression when heterozygous with other pl alleles than as a homozygote (Hollick and Chandler 1998). However, in a study of the maize regulatory locus r1, a correlation between epigenetic modifications and inbreeding depression or heterosis was not observed (Auger et al. 2004).
We have previously shown that the stable WR expression pattern of wild-type P1-wr can be perturbed by a trans-acting epigenetic modifier Ufo1. P1-wr Ufo1 plants produce a red pericarp/red cob glume (RR) phenotype (Chopra et al. 2003). In the current study, the interaction of Ufo1 with the P1-wr* epiallele resulted only in gain of cob glume pigmentation (WR pattern). Bisulfite sequencing established that the recovery of cob color in P1-wr* Ufo1 plants is correlated with demethylation of a 168-bp region of p1 intron 2 to a level similar to that of standard P1-wr. Further, the extent of DNA methylation of the 168-bp region directly correlated with p1 transcription (data not shown). These results suggested that the 168-bp region may contain regulatory elements. A search for regulatory elements in the intron 2 region manually, as well as using the PlantCare database (Lescot et al. 2002), identified CAAT boxes and a RY element. Similar cis sequences have been shown to play a role in the regulation of gene expression (Bobb et al. 1997). Generally, regulatory elements are found in the promoter; however, their occurrence in other genic regions has also been documented. In Arabidopsis, whorl-specific regulation of AGAMOUS, a class C floral organ identity gene, has been attributed to enhancer and suppresser elements present in the second intron of the gene (Sieburth and Meyerowitz 1997; Deyholos and Sieburth 2000). Similarly, intron 1 sequences of the Arabidopsis FLOWERING LOCUS C gene are required for its epigenetic repression through vernalization (Sheldon et al. 2006). In both these examples, DNA methylation and/or chromatin modifications of the intronic sequences containing regulatory elements may cause transcriptional silencing (Chan et al. 2005; Sheldon et al. 2006).
The idea that distinct pericarp- and cob-specific elements are associated with the p1 gene sequence was perceived very early on and presented in the “presetting and erasure” model (Schwartz 1982). Indeed, in P1-rr, a single-copy allele of p1, a 1.2-kb sequence present ∼5 kb upstream from the start of transcription contains basal enhancer element(s) shown to function in pericarp and cob glumes (Sidorenko et al. 2000). Further, in P1-rr′, a transgene-induced paramutagenic allele, and in P1-pr (patterned pericarp/light red cob), a natural epiallele, loss of pericarp and cob pigmentation was correlated with hypermethylation of this region. Additionally, recent analysis of a transposon mutagenesis-derived P1-rw (for red pericarp, white cob) allele has identified a cob-specific 386-bp enhancer sequence (Zhang and Peterson 2005). In P1-rr, there are two copies of the 386-bp enhancer; one is part of the 1.2-kb enhancer while the other lies upstream of this region. In P1-wr, however, there is only one copy of the 386-bp sequence that is present within the 1.2-kb homologous region. The bisulfite sequencing results showed that the 386-bp enhancer (region I in Figure 6) is heavily methylated in P1-wr, P1-wr*, and P1-wr* Ufo1 plants. Moreover, the 168-bp intron 2 sequence is hypomethylated in the P1-wr and P1-wr* Ufo1 plants exhibiting cob glume color. These observations further support the idea that the 168-bp sequence may have an independent cob-specific enhancer activity in P1-wr. We propose that the 168-bp putative enhancer region in intron 2 acts either alone or in combination with the distal enhancers. Hypomethylation of all these enhancers in P1-wr Ufo1 plants and correlated enhancement of pericarp and cob glume color (Chopra et al. 2003) suggest that, when functioning together, these enhancers have an additive effect. It is possible that, due to the multicopy gene structure of P1-wr, the 168-bp intron 2 region of the previous gene copy will effectively function as a distal enhancer for the proximal copy. Experiments are in progress to determine if these sequences function as combinatorial enhancers of p1 gene expression in the multicopy P1-wr allele.
The tissue-specific expression pattern generated by P1-wr has previously been attributed to repeat-induced gene silencing through a mechanism that may involve intrachromosomal interactions among repeats (Chopra et al. 1998). In P1-wr, the direct repeats are coordinately methylated. Coordinate methylation also occurs in P1-wr* copies, although the level of methylation is higher than in P1-wr. Although not completely understood, the role of epigenetic mechanisms in regulating expression of tandem repeated endogenous genes or genes with repeated sequences has been investigated in many systems (see reviews by Birchler et al. 2000; Chan et al. 2005). In Arabidopsis, the expression of FWA is developmentally controlled by DNA methylation of two direct repeats in the transcribed region. Hypermethylation of the repeats silences the gene in vegetative tissues whereas their demethylation by DEMETER, a DNA glycosylase, leads to expression of FWA in endosperm (Soppe et al. 2000; Kinoshita et al. 2004). In maize, the B-I allele of the b1 locus is regulated by epigenetic modifications of tandem direct repeats that are present 100 kb upstream of the transcription start site (Stam et al. 2002). These repeats have a dual role in b1 regulation: they have enhancer activity and are also required for paramutation of B-I. A better insight into B-I paramutation has been provided by the recent cloning of mop1 (mediator of paramutation1), which encodes an RNA-dependent RNA polymerase that is required for epigenetic modifications of the direct repeats (Alleman et al. 2006). From the current study it appears that the wild-type ufo1 interacts with different regions of p1 and maintains their specific epigenetic states to produce stable tissue-specific expression patterns.
The gain of cob pigmentation in P1-wr* Ufo1 was accompanied with demethylation of the intron 2 region in the CG as well as the CNG context. Since maintenance of CG and CNG methylation requires different genetic components (Chan et al. 2005), this result implies that Ufo1 is not a lesion in any of the known DNA methyltransferases. If that were the case, the loss of either CG or non-CG methylation should have been observed. In fact, this feature of Ufo1 resembles mutations in DECREASE IN DNA METHYLATION1 (DDM1), a SWI2/SNF2-like ATP-dependent chromatin remodeling protein that, when mutated, leads to disruption of the maintenance of both CG and non-CG methylation (reviewed in Rangwala and Richards 2004). Inheritance studies of Ufo1 in a test cross indicate that a gain of cob pigmentation in P1-wr* plants is observed even following removal of Ufo1 by meiotic segregation. Similar examples of independent segregation of the effecting mutations and their resulting phenotypes have also been observed in the case of mop1 and ddm1 (Kakutani et al. 1996; Dorweiler et al. 2000). However, both Ufo1 and mop1 differ from ddm1 in that they do not alter DNA methylation of highly repetitive genomic sequences. These observations allow us to propose that ufo1 (wild type) could be acting upstream of DNA methylation and may be involved in maintaining a specific chromatin state. This study also shows that the interaction of Ufo1 is not only allele specific, but epiallele specific; i.e., the phenotypic modification observed is dependent upon the initial epigenetic state of each allele. Whether a tandemly repeated gene structure is an absolute requirement of Ufo1 interaction remains to be determined. Experiments are in progress to study the interaction of Ufo1 with other p1 alleles that have single-copy gene structure and are either hyper- or hypomethylated. The results of these experiments, together with the cloning of Ufo1, will provide new insight into the action of this unique modifier whose action seems to maintain the tissue-specific expression patterns of p1 alleles.
We thank Benjamin Burr and the Maize Genetics Cooperation Stock Center (Urbana, IL) for recombinant inbred lines and maize genetic stocks. We are thankful to Terry Olson for excellent technical assistance and to Michael Robbins and Dawn Luthe for critical reading of the manuscript. This research was performed under the Hatch projects 3855 and 4154 and supported by a National Science Foundation grant 0416425 to S.C.
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession no. EF165349.
Communicating editor: S. R. Wessler
- Received September 20, 2006.
- Accepted December 5, 2006.
- Copyright © 2007 by the Genetics Society of America