Progressive Loss of DNA Methylation Releases Epigenetic Gene Silencing From a Tandemly Repeated Maize Myb Gene
Rajandeep S. Sekhon, Surinder Chopra

Abstract

Maize pericarp color1 (p1) gene, which regulates phlobaphene biosynthesis in kernel pericarp and cob glumes, offers an excellent genetic system to study tissue-specific gene regulation. A multicopy p1 allele, P1-wr (white pericarp/red cob) is epigenetically regulated. Hypomethylation of P1-wr in the presence of Unstable factor for orange1 (Ufo1), leads to ectopic pigmentation of pericarp and other organs. The Ufo1-induced phenotypes show incomplete penetrance and poor expressivity: gain of pigmentation is observed only in a subset of plants carrying Ufo1 mutation, and the extent of pigmentation is highly variable. We show that Ufo1 induces progressive hypomethylation of P1-wr repeats over generations. After five generations of exposure to Ufo1, a 30–40% decrease in CG and CNG methylation was observed in an upstream enhancer and an intron region of P1-wr. Interestingly, such hypomethylation correlated with an increase in penetrance of the Ufo1-induced pigmentation phenotype from ∼27 to 61%. Expressivity of the Ufo1-induced phenotype also improved markedly as indicated by increased uniformity of pericarp pigmentation in the later generations. Furthermore, the poor expressivity of the Uo1 is associated with mosaic methylation patterns of the P1-wr upstream enhancer in individual cells and distinct P1-wr gene copies. Finally, comparison of methylation among different tissues indicated that Ufo1 induces rapid CG and CNG hypomethylation of P1-wr repeats during plant development. Together, these data indicate that the poor penetrance and expressivity of Ufo1-induced phenotypes is caused by mosaicism of methylation, and progressive mitotic hypomethylation leads to improved meiotic heritability of the mutant phenotype. In duplicated genomes like maize, loss of an epigenetic regulator may produce mosaic patterns due to redundancy of epigenetic regulators and their target sequences. We show here that multiple developmental cycles may be required for complete disruption of suppressed epigenetic states and appearance of heritable phenotypes.

METHYLATION of cytosine residues in DNA is a highly conserved epigenetic mark in most eukaryotic organisms. Historically, DNA methylation has been proposed to be a defense mechanism adopted by genomes against selfish DNA elements (Goll and Bestor 2005). Indeed, transposons and other retroelements are generally methylated and loss of methylation at these elements leads to increased transposition (Miura et al. 2001; Singer et al. 2001; Kato et al. 2003; Zhang et al. 2006). In addition to its role in protecting genome integrity, DNA methylation has also been implicated in the regulation of gene expression (Bird 2002; Zhang et al. 2006; Zilberman et al. 2007). Generally, increased cytosine methylation leads to transcriptional silencing and, conversely, loss of methylation results in increased transcriptional activity (Chan et al. 2005; Goll and Bestor 2005; Zilberman et al. 2007). Plasticity of DNA methylation during the life cycle of an organism, and heritability of methylation patterns over generations are crucial for normal development (Schob and Grossniklaus 2006). Genetic manipulations that alter DNA methylation cause various developmental abnormalities in plants (Finnegan et al. 1996; Ronemus et al. 1996; Cao and Jacobsen 2002), mice (Li et al. 1992; Okano et al. 1999), humans (Hansen et al. 1999; Xu et al. 1999), and fungi (Foss et al. 1993; Malagnac et al. 1997).

Unlike in mammals, where most cytosine methylation marks are erased during an early embryonic stage and reestablished later (reviewed in Okano et al. 1999; Bird 2002), plants perpetuate this modification during gametogenesis and embryogenesis (Bender 2004). Another unique difference between plants and mammals lies in their reproductive biology: while germ and somatic cells do not share a common lineage in mammals, plant germ cells are derived from somatic cells (Richards 2006). This provides a mechanism through which epigenetic marks (including DNA methylation) accumulating during the life cycle of plants can be transmitted to the next generation. Therefore, cytosine methylation modifications provide a new source of genetic diversity in plants.

Although recent work has shed light on the importance of DNA methylation in diverse processes, its role in regulation of tissue-specific gene expression patterns is not clear (Gehring and Henikoff 2007). The maize flavonoid biosynthetic pathway is well suited to address this question because: (1) genes encoding structural enzymes and transcription regulators have been genetically and molecularly characterized; (2) multiple natural alleles and epialleles of these genes have been identified; and (3) many of these alleles produce distinct tissue-specific pigmentation patterns. Biosynthesis of phlobaphenes—brick red flavonoids that accumulate in floral organs including pericarp, tassel and cob glumes, and silk—is regulated by an R2R3 Myb transcription factor encoded by pericarp color1 (p1) (Coe et al. 1988; Styles and Ceska 1989). Remarkably, p1 has >100 natural alleles and epialleles, many of which can be easily distinguished on the basis of their expression in the floral organs (Brink and Styles 1966; Cocciolone et al. 2001). The p1 alleles are identified by their pigmentation in pericarp and cob glumes; e.g., P1-rr produces red pericarp and red cob glumes, whereas P1-wr conditions white pericarp and red cob glumes.

Expression patterns of P1-rr and P1-wr are intriguing; both specify red cob but P1-wr fails to produce any pigmentation in pericarp. The two alleles share high (>99.0%) DNA sequence similarity in their coding, proximal promoter, and upstream promoter regions (Chopra et al. 1998). P1-rr is composed of a single gene unit flanked by a 5.2-kb sequence, whereas P1-wr consists of six or more copies of a complete 12.6-kb gene unit arranged in a head-to-tail fashion. Furthermore, P1-wr is densely methylated in the promoter and coding region as compared to P1-rr. On the basis of these differences and characterization of transgenic plants carrying promoter and coding region of P1-rr and P1-wr, the unique expression pattern of P1-wr has been attributed to epigenetic mechanisms (Cocciolone et al. 2001). The nature of these epigenetic mechanisms, however, remains largely unknown.

We have characterized Unstable factor for orange1 (Ufo1), a trans-acting dominant modifier of P1-wr. P1-wr; Ufo1 plants show ectopic accumulation of phlobaphenes in pericarp and other organs including husk, sheath, silks, and tassel glumes. Ufo1-induced phenotypes are not completely penetrant such that only a subset (∼27%) of F1 progeny (P1-wr; Ufo1) plants shows gain of pericarp pigmentation (Chopra et al. 2003; this study). In addition, Ufo1-induced phenotypes exhibit poor expressivity as the pericarp pigmentation ranges from a red scar at the silk attachment point to uniform dark pigmentation. Such incomplete penetrance and poor expressivity of Ufo1 does not seem to be background dependent as we observed similar levels of penetrance in other genetic backgrounds including W22, T232, and B73 inbred lines (R.S. Sekhon and S. Chopra, unpublished results). Phenotypes produced by P1-wr; Ufo1 plants are meiotically unstable as a subset of the progeny reverts back to a normal P1-wr expression pattern. Interestingly, the gain of phlobaphene pigmentation in pericarp is associated with hypomethylation of P1-wr repeats and, importantly, there is a direct correlation between the extent of hypomethylation and levels of pericarp pigmentation (Chopra et al. 2003).

This study was aimed at understanding the mechanism(s) responsible for incomplete penetrance and poor expressivity of the Ufo1-induced epigenetic regulation in tissue-specific gene expression. We studied the Ufo1 and P1-wr interaction for multiple generations and found that each generation of such interaction leads to gradual increase in penetrance and expressivity of the Ufo1-induced phenotypes. Interestingly, such improvement in penetrance and expressivity was accompanied by progressive loss of P1-wr methylation. Furthermore, detailed profiling of methylation of a regulatory region of P1-wr indicated that poor expressivity of the Ufo1-induced phenotypes is due to somatic mosaicism of methylation. Such mosaicism appears to be contributed by differential P1-wr methylation in individual cells and distinct P1-wr gene copies. Finally, we also observed that the presence of Ufo1 leads to rapid loss of P1-wr methylation during plant development. On the basis of these observations, we propose that the function of wild-type ufo1 is to maintain a stable chromatin state at the P1-wr repeat complex. Extended exposure to Ufo1 leads to cumulative destabilization of the P1-wr epigenetic state over generations resulting in improved penetrance and expressivity of ectopic pigmentation phenotypes. Overall, our study shows that, in highly complex and duplicated genomes like maize, redundancy of epigenetic regulators and their target sequences may hinder the phenotypic effects expected from the loss of an epigenetic regulator. Thus, prolonged absence of such regulatory elements for multiple developmental cycles may be needed to produce noticeable phenotypes.

MATERIALS AND METHODS

Maize stocks:

A maize stock carrying the Ufo1 mutation in the P1-wr background was obtained from Derek Styles, University of Victoria, Canada. This mutation was introgressed into an inbred line 4Co63 (genotype p1-ww c1 r-r) as described previously (Chopra et al. 2003). Herein, the homozygous Ufo1 stock in the 4Co63 background is referred to as p1-ww; Ufo1.

Genetic crosses:

p1-ww; Ufo1 plants were crossed to P1-wr (W23 inbred) to obtain F1 progeny (Figure 2A). For developing the backcross population, F1 plants (P1-wr[W23] × p1-ww; Ufo1) with high levels of Ufo1-induced pigmentation on leaf sheath, husk, and tassel glumes were backcrossed to P1-wr[W23] (Figure 2A). Since gain of pericarp pigmentation is the most reliable marker for the presence of Ufo1, highly expressing BC1F1 ears were selected on the basis of uniform dark orange pericarp pigmentation. The BC1F1 plants grown from these selected ears were subjected to subsequent cycles of backcrossing and selection to develop the later populations. To develop the testcross progeny, BC6F1 plants showing Ufo1-induced uniform pericarp pigmentation phenotype were crossed to the p1-ww[4Co63] inbred line and progeny plants were scored for pericarp pigmentation (Figure 2B). Homozygosity of the BC6F1 plants for the P1-wr allele was established by a PCR assay that differentiates between P1-wr and p1-ww[4Co63] alleles (not shown). All the populations discussed here were grown at the Agronomy Farm, Pennsylvania State University, and in the winter nursery at Puerto Rico.

DNA and RNA purification and gel blot hybridizations:

Plant genomic DNA was isolated using a modified CTAB method (Saghai-Maroof et al. 1984). For comparison of DNA methylation among different leaves, DNA was isolated from alternate leaves of a plant at the time of tassel emergence. For all other experiments, the fifth leaf from the base of plants at 5- to 6-leaf stage was harvested for DNA extraction. Restriction enzyme digestions were performed using enzymes, reagents, and incubation conditions from Promega (Madison, WI). The restricted DNA was fractionated on 0.8% agarose gels, transferred to nylon membranes, and the gel blots were sequentially hybridized with α32P-dCTP labeled DNA fragments. For RNA gel blot analysis, total RNA was isolated from various tissues using lithium chloride buffer [100 mm lithium chloride, 100 mm Tris HCl (p.H. 8.0), 10 mm EDTA, 1% SDS] and precipitated with 4 m lithium chloride (Barlow et al. 1963). Total RNA was fractionated on denaturing gels, blotted onto nylon membranes, and hybridized with α32P-dCTP probe fragments as described previously (Boddu et al. 2004). DNA fragments F15 and F8B belonging to an upstream regulatory region and second intron, respectively, were used as probes in Southern blot analysis (Lechelt et al. 1989; Grotewold et al. 1991; Sekhon et al. 2007). Probe fragments used for Northern analysis were obtained from the plasmids pC2 containing a maize C2 cDNA (Wienand et al. 1986), pChi1 containing a maize Chi1 cDNA (Grotewold and Peterson 1994), and pZmA1 with a maize DFR cDNA (Schwarz-Sommer et al. 1987). Plasmid DNA was prepared using Strataprep plasmid miniprep kit (Stratagene, La Jolla, CA) following the manufacturer's directions. The Southern and Northern blots were stripped by washing for 15 min in boiling solution of 0.1% SDS before rehybridization.

Quantification of flavan-4-ols:

Flavan-4-ols are colorless precursor compounds that polymerize to form red phlobaphene pigments (Styles and Ceska 1977). These compounds have absorption maxima of 564 nm. Relative accumulation of flavan-4-ols was estimated by spectral analysis following previously published method (Waterson and Butler 1983; Grotewold et al. 1998). Briefly, pericarp and cob tissues were collected at 20 days after pollination and immediately frozen in liquid nitrogen. Two hundred milligrams of tissue from each genotype was ground in a mortar with a pestle and transferred to a 15-ml centrifuge tube containing 1 ml of 30% HCl/70% butanol (v/v) solution. Samples were incubated for 1 hr at 37°, spun at 14000 × g for 10 min, and the supernatant was collected in a 2-ml tube. Absorbance was measured using UV Mini 1240 spectrophotometer (Shimadzu Scientific Instruments, Columbia, MD). The absorbance value of these extracts at 564 nm (λmax) was used as an estimate of flavan-4-ols. Two independent samples from one plant were subjected to this analysis and mean and standard deviation of the replicates is presented. Flavan-4-ols are heat labile and boiling leads to elimination of the 564-nm peak that is diagnostic of these compounds (Grotewold et al. 1998). Thus, the extracts were boiled for 10 min and spectral measurements of the boiled samples were recorded to confirm elimination of the 564-nm peak.

Genomic bisulfite sequencing:

DNA methylation of the upper DNA strand from a 333-bp upstream promoter region and a 312-bp intron 2 region of P1-wr was assayed by genomic bisulfite sequencing. Eight micrograms of leaf genomic DNA was completely digested with suitable restriction enzymes that cut outside the regions of interest. The digested DNA was treated with sodium bisulfite following a previously published protocol (Jacobsen et al. 2000) with minor modifications (Sekhon et al. 2007). Regions of interest were PCR amplified using P1-wr primers specifically designed to amplify DNA modified with sodium bisulfite (supplemental Table 1). The resulting PCR products were gel purified, cloned using TOPO TA cloning kit (Invitrogen, Carlsbad, CA), and 20–30 clones/ligation/genotype were sequenced to obtain the bisulfite modified sequence data.

RESULTS

Penetrance of Ufo1-induced phenotypes correlates with p1-regulated phlobaphene biosynthesis:

P1-wr × Ufo1 interaction results in a gain of pigmentation in pericarps and cob glumes of a subset of progeny plants (Chopra et al. 2003), and pigmentation levels in individual plants are quite variable (Figure 1A). We tested whether such phenotypic differences are due to differential expression of p1-regulated pathway genes and/or relative levels of flavan-4-ols, compounds that polymerize to produce phlobaphenes. While only trace amounts of flavan-4-ols were present in pericarp of F1 (P1-wr/p1-ww; Ufo1/ufo1) plants that did not show pericarp pigmentation (hereafter called nonexpressers), substantial levels were detected in sibling plants showing Ufo1-induced gain of pericarp pigmentation (hereafter called expressers) (Figure 1B). Steady state transcript levels of three p1-regulated structural genes c2, chi1, and a1 were compared in various floral and vegetative tissues of sibling F1 plants. In nonexpressers, only silk tissue had appreciable amounts of transcripts of the three structural genes (Figure 1C). Conversely, the expressers had a significant increase in transcript accumulation of these genes in pericarp, silk, and husk. Overall, these results indicate that phenotypic differences among sibling expresser and nonexpresser plants are a result of differential activity of the p1-regulated phlobaphene biosynthetic pathway.

Figure 1.—

Ufo1-induced upregulation of phlobaphene biosynthesis in floral and vegetative organs. (A) Gain of pericarp and cob pigmentation in the F1 (P1-wr/P1-wr; ufo1/ufo1 × p1-ww/p1-ww; Ufo1/Ufo1) individual plants. Shown here are representative ears from sibling F1 plants that either lack pericarp pigmentation (nonexpressers, N-1 and N-2), or show Ufo1-induced pericarp pigmentation (expressers, U-1 and U-2). (B) Pericarp and cob pigmentation levels of ears in A correlate with accumulation of flavan-4-ols. Levels of flavan-4-ols are based on relative absorbance (564 nm) of extracts from tissues of individual plants (see materials and methods). (C) Steady-state transcript levels in floral and vegetative tissues. Three p1-regulated structural genes of the phlobaphene biosynthetic pathway were assayed. Gel blot carrying total RNA isolated from different tissues of two sibling plants shown in A was sequentially hybridized with probes of genes shown on the left.

Inheritance of Ufo1-induced pigmentation phenotype improves over generations:

We have previously shown that Ufo1 exhibits incomplete penetrance and poor expressivity (Chopra et al. 2003). Ufo1 induces varying degrees of hypomethylation from P1-wr repeats and that the extent of hypomethylation is directly correlated with the level of pericarp pigmentation (Chopra et al. 2003). Thus, it is possible that the variability of P1-wr expression due to such differential methylation patterns is manifested as poor expressivity and, eventually, incomplete penetrance of pigmentation phenotype. Furthermore, Ufo1 appears to be a weak modifier that is not able to induce complete hypomethylation of P1-wr. However, in such a scenario, prolonged exposure of P1-wr to Ufo1 might cause increased destabilization of the P1-wr epigenetic state and lead to improved heritability of the pericarp pigmentation phenotype. To test this idea, we looked at penetrance of Ufo1 in F1, five backcross populations, and a testcross population (see Figure 2, A and B for details). In the F1 generation, ∼27% of P1-wr; Ufo1 plants showed gain of pericarp pigmentation instead of 100% expected from a dominant monogenic trait (Figure 2C). Additionally, the Ufo1-expressing F1 plants exhibited a range of pericarp pigmentation indicating poor expressivity (Figure 2D). Remarkably, the penetrance of Ufo1 showed an incremental improvement after each backcross (Figure 2C). For example, of 983 BC5F1 plants expected to show Ufo1-induced gain of pericarp pigmentation, 600 exhibited the phenotype indicating 61% penetrance in this generation. Expressivity of Ufo1-induced phenotypes also improved upon backcrossing as evident from relatively uniform pericarp pigmentation in the BC5F1 (Figure 2D). To further confirm if such improved penetrance and expressivity persists in later generations, we testcrossed strong expressers of BC6F1 generation with p1-ww[4Co63] (Figure 2B). Interestingly, of 127 plants expected to carry Ufo1 mutation, 82 showed uniform pericarp pigmentation, indicating 65% penetrance of Ufo1 in this population. In conclusion, these results indicate that extended exposure of P1-wr to Ufo1 leads to improved inheritance of pericarp pigmentation phenotype.

Figure 2.—

Penetrance and expressivity of gain of pericarp pigmentation phenotype improves upon extended exposure of P1-wr to Ufo1 for multiple generations. (A and B) Crossing scheme used to develop the F1 and BC5F1 and BC6 testcross (BC6 TC) progenies. (C) Penetrance of Ufo1 in different populations. Expected phenotypes are based on a hypothetical assumption that Ufo1 has complete penetrance. In the BCnF1 population, 50% of the progeny are expected to carry Ufo1 (genotype: P1-wr/- ; Ufo1/ufo1) and show Ufo1-induced gain of pericarp pigmentation. In the BC6 TC population, all the individuals will be heterozygous at the p1 locus (P1-wr/p1-ww) and show 1:1 segregation for the ufo1 locus (Ufo1/ufo1:ufo1/ufo1). Thus, 50% of the BC6 TC individuals are expected to show gain of pericarp pigmentation. Penetrance was calculated as percentage of observed/expected Ufo1-expressers. (D) Pericarp and cob pigmentation in representative F1 and BC5F1 progeny plants.

Ufo1 causes progressive loss of DNA methylation of P1-wr over generations:

Ufo1-induced gain of pigmentation is associated with decreased DNA methylation of P1-wr repeats, and such hypomethylation in individual P1-wr; Ufo1 plants is variable and shows a direct correlation with level of pericarp pigmentation (Chopra et al. 2003; Sekhon et al. 2007). We tested if improved penetrance and expressivity of gain of pigmentation phenotypes after multiple generations of exposure to Ufo1 could be due to progressive hypomethylation of the P1-wr multicopy complex. DNA methylation of F1 and BC4F1 plants was compared by digesting leaf genomic DNA with HpaII and hybridizing the gel blot with a p1 fragment 15 as a probe (Figure 3A). In P1-wr, probe 15 hybridized to a prominent band of ∼7.9 kb, a weak band of ∼12.0 kb, and three smaller size bands of ∼0.6, 0.5, and 0.4 kb (Figure 3B). In F1 expresser plants, three new fragments of ∼3.6-, 3.0-, and 2.3-kb size were observed, indicating hypomethylation of additional HpaII sites. In BC4F1, interestingly, the number and intensity of HpaII bands increased: a total of seven bands were observed in the Ufo1-expressing individuals and intensity of these bands was higher than that observed in the F1 expressing plants. The low molecular weight bands were either absent or very light in the nonexpresser plants. These results suggested that P1-wr repeat complex is more hypomethylated in BC4F1 as compared to F1 generation.

Figure 3.—

Progressive loss of DNA methylation of P1-wr in the presence of Ufo1. (A) Gene structure and methylation map of P1-wr. Solid line represents a complete 12,596-bp single P1-wr repeat. Transcription start site is indicated by a bent arrow. Boxes superimposed on the solid line represent three exons marked as E1, E2, E3; blank regions of E1 and E3 represent UTRs. Probe fragments are shown at the top of the gene structure. Striped boxes at the bottom of the gene structure represent two regions subjected to genomic bisulfite sequencing (see results). Methylation map of HpaII sites in p1 sequence of P1-wr and P1-wr; Ufo1 plants is shown at the bottom of the gene structure. Open, striped, and solid hexagons represent unmethylated, partially methylated, and completely methylated HpaII sites, respectively. (B) Gel blots showing DNA methylation differences between F1 and BC4F1 plants (see Figure 2 for scheme used to develop these populations). Leaf genomic DNA of respective plants was digested with HpaII and gel blots were hybridized with p1 probe fragment 15. Abbreviations are: N, Ufo1 nonexpressing plants carrying wild-type P1-wr (phenotype: white pericarp, red cob); W, plants carrying null p1-ww allele (phenotype: white pericarp, white cob); U, Ufo1-expressing plants (phenotype: red pericarp, red cob). Molecular weight marker bands are shown in kilobase pairs (kb) on right. Bands hybridizing with the probe are indicated by arrows on the right side along with their sizes in kilobases.

To determine the exact position and context of cytosines undergoing Ufo1-induced methylation modifications, we performed genomic bisulfite sequencing. Two regions were chosen for this assay: a 333-bp fragment from a distal enhancer region, and a 312-bp region from the 5′ end of intron 2 of P1-wr (see Figure 3A for location of these regions). Both regions undergo significant methylation modifications during P1-wr × Ufo1 interaction and have been considered important for upregulation of P1-wr expression (Chopra et al. 2003; Sekhon et al. 2007; Robbins et al. 2008). In the distal enhancer region, overall CG and CNG methylation was very high (∼95%) in P1-wr as well as in the F1 expressers (Figure 4A). Expresser plants in the BC4F1 generation, however, had significantly lower levels of CG (64%) and CNG (50%) methylation. Interestingly, CHH methylation levels were very low in all three genotypes. Individual CG and CNG sites also showed a similar trend; methylation levels were comparable for P1-wr and F1 plants but showed sizeable decrease in the BC4F1 generation (supplemental Figure 1, A and B). In summary, while Ufo1 did not induce noticeable changes in CG and CNG methylation of the P1-wr distal enhancer region in F1 plants, significant hypomethylation was observed in the BC4F1 generation.

Figure 4.—

Bisulfite data showing Ufo1-induced progressive hypomethylation of two distinct regions of P1-wr. Cumulative methylation in CG, CNG (N is any nucleotide), and CHH (H is A, C, or T) context in a 333-bp distal enhancer (A), and a 312-bp intron 2 region (B) is shown (see Figure 2A for location of these regions). The genotypes assayed were wild-type W23 carrying the homozygous P1-wr allele (designated as P1-wr), and F1 and BC4F1 progenies showing Ufo1-induced pericarp pigmentation phenotype (see Figure 2 for crossing scheme used to develop these populations). For each genotype, overall methylation in each context was calculated by dividing the number of methylated cytosines by total number of cytosines in the context in all the clones.

Likewise, in the intron 2 region, cumulative CG and CNG methylation levels did not show striking differences between P1-wr and F1 plants. CG methylation was high in P1-wr (96%) as compared to F1 (82%) (Figure 4B), while CNG methylation in F1 (81%) was more than that in P1-wr (73%). In BC4F1, however, CG and CNG methylation levels dramatically decreased to 48% and 40%, respectively. CHH methylation levels in the intron 2 region were very low in all three genotypes. The methylation pattern of individual CG sites resembled that of overall methylation in this region; methylation of most CG sites was highest in P1-wr, slightly decreased in F1, and was lowest in the BC4F1 generation (supplemental Figure 2A). Methylation of individual CNG sites was more variable; while some sites showed hypermethylation in the F1 generation, others were hypomethylated as compared to P1-wr (supplemental Figure 2B). In the BC4F1 generation, however, all CNG sites were less methylated as compared to P1-wr and F1 plants. Overall, combined data from DNA gel blot analysis and genomic bisulfite sequencing indicated that long-term exposure of Ufo1 leads to increased P1-wr hypomethylation in the distal enhancer and second intron.

Ufo1 induces rapid hypomethylation of P1-wr during plant development:

In the presence of Ufo1, decrease of P1-wr methylation from the F1 to the BC4F1 generation indicated that the hypomethylated state was meiotically transmitted. To test if Ufo1-induced P1-wr hypomethylation progresses mitotically during plant development, we compared methylation of different leaves. Leaf genomic DNA of expresser and nonexpresser BC5F1 plants was digested with HpaII and the resulting gel blot was probed with the p1 distal enhancer fragment 15. Ufo1-induced hypomethylation of P1-wr in these plants is shown by the presence of three fragments (indicated by arrows in Figure 5A). Interestingly, intensity of these bands varied among different leaves of the expresser plant: while barely detectable in first, third, and fifth leaves, these bands had higher signal intensity in the seventh, ninth, and eleventh leaves. Similarly, hybridization of the same blot with probe 8B (intron 2 region) also showed incremental loss of methylation (compare the intensity of two bands marked by arrows in Figure 5B). Thus, in the expresser plants, the loss of methylation progresses from the bottom leaf to the terminal leaf. These data indicated that P1-wr exhibits developmental hypomethylation in the presence of Ufo1. The small-size bands indicative of Ufo1-induced P1-wr hypomethylation were not observed in nonexpressers (data not shown) suggesting an absence of developmental P1-wr hypomethylation in the absence of Ufo1.

Figure 5.—

Mitotic progression of Ufo1-induced hypomethylation of P1-wr during plant development. Shown here is a comparative methylation pattern of leaves at different developmental stages. Genomic DNA from different leaves of a BC5F1 expresser plant was digested with HpaII and hybridized with a p1 promoter fragment 15 (A) and an intron fragment 8B (B). Numbers on the top of lanes represent leaves, 1 being the first leaf and 11 being the flag leaf. Molecular weight marker bands in kilobase pairs are shown in the middle of the two blots.

To further confirm the developmental hypomethylation of P1-wr sequence and to identify the sequence context undergoing such modifications, bisulfite sequencing data was compared between the fifth leaf and pericarp, two organs representing distinct developmental stages. Since the Ufo1-induced phenotypic changes are clearly evident in pericarp, and the fifth leaf is a nonexpressing tissue for p1, the developmental hypomethylation is expected to be higher in pericarp tissue. For this comparison, we chose the 333-bp region from the distal enhancer (see Figure 3A for location). Bisulfite assay showed that, even in the wild-type P1-wr, this region had less CG and CNG methylation in pericarp as compared to leaf, indicating hypomethylation during development (Figure 6). This result is consistent with the fact that, although the P1-wr allele does not induce phlobaphene pigmentation in the pericarp, p1 transcripts do accumulate in this tissue albeit at lower levels as compared to cob glumes (Chopra et al. 1996). Interestingly, in the F1 expresser plants, the hypomethylation in CG and CNG context was more drastic in pericarp as compared to leaf. In summary, these data show that DNA methylation decreases during differentiation of cells from the fifth leaf, an organ that represents an early developmental stage and lacks detectable p1 expression, to pericarp tissue, which represents a later stage and accumulates detectable levels of p1 transcripts. Furthermore, presence of Ufo1 causes a rapid change in the epigenetic state of P1-wr during development, which is consistent with increased p1 transcription in pericarp and a correlative gain of phlobaphene pigmentation (Chopra et al. 2003).

Figure 6.—

P1-wr methylation differences among tissues belonging to vegetative and reproductive phases of development. Methylation of a 333-bp distal enhancer region of P1-wr (see Figure 2A for location) in fifth leaf and pericarp was assayed by genomic bisulfite sequencing. Genotypes compared were homozygous P1-wr/P1-wr; ufo1/ufo1 (designated as P1-wr; ufo1) and an F1 generation of the cross P1-wr/P1-wr; ufo1/ufo1 × p1-ww/p1-ww; Ufo1/Ufo1 (designated as P1-wr; Ufo1). Calculations used to make this figure and other details are the same as for Figure 4.

Ufo1 induces increased mosaicism of P1-wr methylation:

Variable somatic expression and poor penetrance of Ufo1-induced pericarp pigmentation could be due to differential methylation patterns of P1-wr in different gene copies and/or cells. Such mosaicism of P1-wr methylation may affect expression in individual cells thereby producing uniformly pigmented, variegated, or nonpigmented pericarp phenotypes. Additionally, the methylation state of P1-wr in the cell contributing to gamete formation may decide penetrance and thus the pericarp pigmentation of the progeny. Analysis of CG and CNG sites in individual clones of the 333-bp distal enhancer region of pericarp of P1-wr and F1 expresser plants revealed a very interesting methylation pattern (Figure 7). Of 24 clones derived from P1-wr pericarps, 1 was hypomethylated, 2 showed a mosaic pattern characterized by a mixture of methylated and nonmethylated sites, while 21 clones were highly methylated. An interesting change in methylation pattern was observed in F1 pericarps; of 26 clones examined, 9 were hypomethylated, 7 showed a mosaic pattern, while 10 were hypermethylated. These results show that the presence of Ufo1 leads to substantial increase in the number of P1-wr DNA molecules with partial or complete hypomethylation. It should be noted that such hypomethylated clones could be contributed by different P1-wr copies, individual cells, or both. The presence of some indels (see positions −4984, −4977, −4900, and −4899 in clones RBS287 and RBS286, and positions −4900 and −4899 in clones RBS300 and RBS 281) indicates that these clones are derived from different copies. To summarize, P1-wr; Ufo1 plants show enhanced methylation mosaicism that is contributed by differential P1-wr methylation in individual cells and distinct gene copies.

Figure 7.—

Ufo1-induced somatic variability of P1-wr methylation at CG and CNG sites in pericarp tissue. Shown here is the methylation state of CG (ovals) and CNG (squares) sites in individual clones obtained by cloning a 333-bp distal enhancer region of P1-wr. Genotypes compared were homozygous P1-wr/P1-wr; ufo1/ufo1 (designated as P1-wr; ufo1) and an F1 generation of the cross P1-wr/P1-wr; ufo1/ufo1 × p1-ww/p1-ww; Ufo1/Ufo1 showing Ufo1-induced pericarp pigmentation (designated as P1-wr; Ufo1). Gene structure of P1-wr is shown at the top and represents a complete 12,596-bp single P1-wr repeat. Transcription start site is indicated by a bent arrow. Boxes superimposed on the solid line represent three exons marked as E1, E2, E3; white portions of E1 and E3 represent UTRs. Probe fragment 15 is shown at the top of the gene structure. The striped box at the bottom of the gene structure represents the 333-bp distal enhancer region subjected to genomic bisulfite sequencing; coordinates of the region and density of CG and CNG sites is shown in the blown-up image. The first row (indicated by black ovals and squares) represents the position of CG and CNG sites in the region, respectively. Clones with all the CG and CNG sites lacking methylation are designated as unmethylated while those with most of these sites methylated are referred to as hypermethylated. Mosaic clones are those having a mixture of methylated and unmethylated CG and CNG sites. Absence of some CG and CNG sites in some of the cells indicates the presence of sequence polymorphisms among P1-wr copies.

DISCUSSION

The white pericarp/red cob phenotype of P1-wr has been attributed to the hypermethylated state of its tandemly repeated copies (Chopra et al. 1998; Cocciolone et al. 2001). The presence of Ufo1 induces varying levels of hypomethylation of the P1-wr repeat complex and a corresponding gain of pericarp pigmentation. However, the extent of Ufo1-induced pigmentation is highly variable and shows poor penetrance. In maize, such poor penetrance has been attributed to the redundancy of the genes controlling/contributing to the phenotype (Sato et al. 1999; Vega et al. 2002). Furthermore, epigenetic states of eukaryotic genes are collectively maintained by DNA methylation and chromatin modifications, two redundant and mutually dependent epigenetic marks that are usually imposed by RNA-based mechanisms (reviewed in Bender 2004; Tariq and Paszkowski 2004; Klose and Bird 2006). Due to redundancy, loss of one of these components may partially be compensated by the other and could, therefore, lower the severity of mutant phenotype. However, loss of one of the components may produce small changes in epigenetic states that are cumulative over generations. This indeed is evident from studies involving mop1 (mediator of paramutation1), an RNA-dependent RNA polymerase (RdRP) in maize, and DDM1 (DECREASE IN DNA METHYLATION1), a SWI2/SNF2-like protein in Arabidopsis (Jeddeloh et al. 1999; Alleman et al. 2006). Mutations in these genes lead to slow but cumulative “decay” of epigenetic states of their target sequences and it takes several generations before phenotypic and molecular change are observed (Vongs et al. 1993; Kakutani 1997; Woodhouse et al. 2006). As proposed in these studies, progressive improvement in penetrance of Ufo1 in this study may also be due to redundancy epigenetic marks.

Partial maintenance of epigenetic state over multiple mitotic divisions can generate somatic cell lineages with either silent or active gene expression state and thus produce variegation patterns (reviewed in Richards 2006). Cellular mosaicism and range of DNA methylation at the population level is manifested as poor expressivity and penetrance, respectively (Matzke and Matzke 1991; 1998; Mcgowan et al. 1989). Herein, we present the evidence that the poor expressivity of Ufo1-induced phenotypes could be due to a strikingly distinct methylation profile of a P1-wr regulatory region in individual pericarp cells and gene copies (Figure 7). Depending upon the P1-wr epigenetic state in the somatic lineages, the pericarps may be uniformly pigmented, variegated, or colorless. This line of thinking is supported by the observation that P1-wr; Ufo1 plants show a range of methylation that correlates with the extent of pericarp pigmentation (Chopra et al. 2003; Sekhon et al. 2007). Thus our study shows that, in tandemly duplicated genes like P1-wr, the differential epigenetic state of regulatory regions of individual gene copies can also add to mosaicism of epigenetic states. In complex and redundant genomes like maize wherein approximately one-third of the genes are tandemly duplicated (Messing et al. 2004), this adds another layer of complexity in epigenetic regulation of gene expression.

Epigenetic states change during the development of an organism as reported for Pl-Blotched allele of purple plant1 (pl1) gene and Robertson's Mutator elements in maize (Martienssen et al. 1990; Hoekenga et al. 2000). It is known that such fine tuning of epigenetic states regulates development- and tissue-specific gene expression in higher eukaryotes (Poethig 1990; Lu et al. 2008). Indeed, comparison of leaf and pericarp methylation showed that P1-wr expression is developmentally regulated (Figure 6). The leaf used for this comparison represents an adult vegetative stage while the pericarp is associated with reproductive phase (Poethig 1990; 2003). We found that such a developmental change from leaf to pericarp is associated with loss of CG and CNG methylation. This observation is consistent with the P1-wr expression pattern; while leaf does not accumulate detectable p1 mRNA, appreciable levels of transcripts are found in the pericarp (Chopra et al. 1996). However, pericarps of P1-wr are not pigmented possibly because a threshold level of p1 mRNA required for activation of downstream genes and thus phlobaphene pigmentation is not achieved (Chopra et al. 2003). In the presence of Ufo1 mutation, there is a rapid destabilization of P1-wr epigenetic states as indicated by reduced CG and CNG methylation. This observation is consistent with increased transcription of p1 and p1-regulated genes and the appearance of pericarp pigmentation in P1-wr; Ufo1 plants.

In plants, DNA methylation modifications accumulating during the life cycle can be transmitted to the next generation because plant germ cells are derived from somatic cells. In maize, the megaspore mother cell (MMC) is derived from the pericarp precursor cells, although the nature of cell lineages that lead to MMC formation is debatable (Greenblatt 1985; Zhang 1999). Depending upon the methylation state of P1-wr in the pericarp cell involved in gamete formation, any of the methylation patterns can be represented in the egg cell and subsequently in the progeny (Figure 7). Increased hypomethylation of the P1-wr complex due to extended exposure to Ufo1 may result in higher numbers of cells carrying hypomethylated P1-wr. This in turn may lead to an increased chance of gametes being derived from cells with such a hypomethylated state of P1-wr resulting in improved inheritance of the pigmented phenotype. Furthermore, it may also reduce the disparity of epigenetic states of P1-wr in the pericarp cells and gametes, thereby contributing to enhanced meiotic inheritance of the mutant phenotype.

Since recurrent selection of individuals with intense and uniform pigmentation was performed, one could argue that such selection lead to accumulation of modifiers resulting in increased penetrance and expressivity. Indeed, in maize, poor penetrance of certain mutant phenotypes has also been attributed to the suppressors/enhancers present in the background (Sato et al. 1999; Vega et al. 2002). Therefore, the improved penetrance of the Ufo1-induced pericarp pigmentation phenotype in BC5F1 as compared to the F1 population could be due to accumulation of favorable modifiers. However, a testcross of BC6F1 should produce individuals with genetic makeup as well as phenotypic penetrance identical to the F1 population. Since the testcross progeny also showed high (∼65%) penetrance of Ufo1-induced pigmentation phenotype, it seems that the increased penetrance here is not solely due to accumulation of modifiers.

Ufo1 mutation has not been cloned and thus the molecular nature of this modifier remains elusive. On the basis of this study and past studies (Chopra et al. 2003; Sekhon et al. 2007; Robbins et al. 2008), it seems that a function of ufo1 (wild type) is to maintain a tissue-specific expression pattern of P1-wr by modulating its chromatin state during development. Disruption of the P1-wr chromatin state in Ufo1 (mutant) background results in ectopic expression in pericarp and other tissues. Once the epigenetic state of the P1-wr locus is destabilized in the Ufo1 mutant background, it may either go back to its normal state and produce wild-type phenotype or maintain the altered expression pattern. Selection of plants showing pericarp pigmentation phenotype helps in perpetuation of the altered P1-wr chromatin state and the absence of the ufo1 function in these plants causes further destabilization of the epigenetic state over generations. To test this hypothesis, the chromatin state of the P1-wr repeat complex needs to be studied through chromatin immunoprecipitation analysis.

Acknowledgments

We thank David Braun and Michael Robbins for helpful suggestions and critical reading of the manuscript. This work was supported by a National Science Foundation award MCB-0619330, a Hatch 4154 award to S.C., and dissertation award from the College of Agricultural Sciences and Research Assistantships from the Department of Crop and Soil Sciences and Plant Biology Graduate Program (to R.S.S.).

Footnotes

  • 1 Present address: Great Lakes Bioenergy Research Center, University of Wisconsin, Madison, WI 53706.

  • Communicating editor: D. Voytas

  • Received October 6, 2008.
  • Accepted November 7, 2008.

References

View Abstract