Abstract
We have characterized Unstable factor for orange1 (Ufo1), a dominant, allele-specific modifier of expression of the maize pericarp color1 (p1) gene. The p1 gene encodes an Myb-homologous transcriptional activator of genes required for biosynthesis of red phlobaphene pigments. The P1-wr allele specifies colorless kernel pericarp and red cobs, whereas Ufo1 modifies P1-wr expression to confer pigmentation in kernel pericarp, as well as vegetative tissues, which normally do not accumulate significant amounts of phlobaphene pigments. In the presence of Ufo1, P1-wr transcript levels and transcription rate are increased in kernel pericarp. The P1-wr allele contains approximately six p1 gene copies present in a hypermethylated and multicopy tandem array. In P1-wr Ufo1 plants, methylation of P1-wr DNA sequences is reduced, whereas the methylation state of other repetitive genomic sequences was not detectably affected. The phenotypes produced by the interaction of P1-wr and Ufo1 are unstable, exhibiting somatic mosaicism and variable penetrance. Moreover, the changes in P1-wr expression and methylation are not heritable: meiotic segregants that lack Ufo1 revert to the normal P1-wr expression and methylation patterns. These results demonstrate the existence of a class of modifiers of gene expression whose effects are associated with transient changes in DNA methylation of specific loci.
PLANT genes involved in pigment biosynthetic pathways have been highly suitable for genetic studies because of the readily visible phenotypes (reviewed in Winkel-Shirley 2001). In maize, the synthesis of two broad categories of flavonoid pigments—anthocyanins and phlobaphenes—is controlled by a well-characterized set of regulatory and structural genes (Styles and Ceska 1977, 1989; Ludwiget al. 1990; Grotewoldet al. 1991). The pericarp color1 (p1) gene controls the synthesis of flavan-4-ol and phlobaphenes. The p1 gene encodes a R2R3 MYB domain protein and directly regulates the transcription of structural genes required for phlobaphene biosynthesis (Grotewold et al. 1991, 1994). The pigmentation phenotypes specified by p1 are most obvious in the kernel pericarp (seed coat) and cob glumes of mature maize ears. Many allelic forms of p1 have been recognized and classified on the basis of their tissue-specific pigmentation (Brink and Styles 1966). Two well-characterized alleles that differ strikingly in their pericarp phenotype are P1-wr (white pericarp and red cob) and P1-rr (red pericarp and red cob). The phenotypic differences between P1-wr and P1-rr have been attributed to their differential transcriptional regulation (Chopraet al. 1996), which in turn may be a function of their unique gene structures. The P1-rr allele carries a single coding sequence (Grotewoldet al. 1991) flanked by 5.2-kb direct repeat sequences, which contain regulatory elements (Sidorenkoet al. 2000; Coccioloneet al. 2001). In contrast, P1-wr contains six copies of a 12.6-kb sequence containing the coding and presumptive P1-wr regulatory regions. The six P1-wr copies are arranged in direct orientation as a multicopy tandem repeat complex. Additionally, the P1-wr gene copies are hypermethylated in their coding and noncoding sequences regions relative to the P1-rr allele. On the basis of these observations, we previously proposed that the P1-wr multicopy structure may result in intraallelic interactions that give rise to silencing of P1-wr expression in kernel pericarp (Chopraet al. 1998).
Functional analysis of P1-wr promoter and coding sequences in transgenic maize plants, as well as studies of natural p1 variants, have provided further support for the hypothesis that the organ-specific expression pattern of P1-wr is epigenetically regulated (Coccioloneet al. 2001). Other studies on regulation of maize flavonoid biosynthetic pathways have also led to the identification of epigenetic regulatory mechanisms (Das and Messing 1994; Dorweileret al. 2000; Hoekengaet al. 2000; Hollicket al. 2000; Hollick and Chandler 2001). In plants, changes in epigenetic states are not associated with changes in DNA sequence, but are accompanied by alterations in DNA methylation (reviewed in Martienssen and Richards 1995; Kooteret al. 1999). Transcriptional gene silencing has been associated with increased DNA methylation and local chromatin compaction (Das and Messing 1994; Lundet al. 1995; Ye and Singer 1996). Additionally, homology-dependent gene silencing (HDGS) of plant transgenes has been associated with the presence of multiple transgene copies (reviewed in Flavell 1994; Matzkeet al. 1996; Kooteret al. 1999). Phenomena similar to HDGS have been found in some cases of endogenous genes with multiple copies in Arabidopsis (Bender and Fink 1995), soybean (Todd and Vodkin 1996), and maize (Ronchiet al. 1995).
We report here the characterization of Ufo1, a factor that, in the presence of a P1-wr allele, induces striking kernel pericarp and plant pigmentation. The Ufo1 factor by itself does not induce pigmentation, nor does it exhibit any detectable effects with alleles other than P1-wr. We show that Ufo1 increases the levels of P1-wr transcripts and the rate of P1-wr transcription in pericarp nuclei. Moreover, P1-wr sequences exhibit reduced levels of DNA methylation in the presence of Ufo1. Interestingly, the activation of P1-wr by Ufo1 is transient: the P1-wr expression and methylation patterns revert to their former state in progeny plants that lack Ufo1. We discuss these results in relation to current models of transcriptional gene silencing.
MATERIALS AND METHODS
Genetic stocks: A stock containing P1-wr and Ufo1 was obtained from Derek Styles, University of Victoria (British Columbia, Canada). This stock was crossed with an inbred line 4Co63 of genotype P1-ww c1 r-r, the F1 was self-pollinated, and F2 progeny plants that were of P1-ww genotype were identified by their colorless tassel glume margins (Zhanget al. 2000). These P1-ww F2 plants were outcrossed to a standard P1-wr inbred line, W23, and 20 plants were grown from each outcross. The resulting progeny plants that carried Ufo1 were identified by orange pigmentation of leaf sheath, husk, and kernel pericarp. Subsequently, the Ufo1 stock was maintained by repeated backcrossing to the W23 inbred line. The inbred line W23 (genotype P1-wr c1 r-g) and other inbred lines C123, B73, and W220 were obtained from the Maize Genetics Cooperation Stock Center (Urbana, IL). The P1-ww [4Co63] was obtained from the National Seed Storage Laboratory (Fort Collins, CO). The P1-rr4B2 and P1-ww1112 alleles used in this study have been previously described (Athma and Peterson 1991), as has allele P1-rr-CSF327 (Coccioloneet al. 2001).
DNA and RNA purification and Northern and Southern hybridization: Plant genomic DNA was prepared using a modified CTAB method (Sagai-Maroofet al. 1984). Restriction enzyme digestions were performed using enzymes, reagents, and incubation conditions from Promega (Madison, WI). Pericarp and cob glumes dissection, RNA extraction, poly(A)+ RNA purification, and gel blotting was done as described previously (Chopraet al. 1996). Gel blots were stripped by washing for 15 min in boiling solution of 0.1% SDS before rehybridization. Plasmid DNA was prepared using the Maxi-prep DNA isolation kit (Promega). DNA fragments of p1 used as probes have been described previously (Lecheltet al. 1989; Chopraet al. 1998; Coccioloneet al. 2001). The extent of genome-wide methylation was determined using repetitive DNA sequences as probes on blots carrying digests of genomic DNA from P1-wr ufo1 and P1-wr Ufo1 plants. These probes include p185 containing maize 185-bp knob repeat sequence (Ananievet al. 1997), pMTY7SC1carrying maize telomeric sequence present near the maize p1 gene on chromosome 1S (Gardineret al. 1996), and pCT4.2 containing 5S ribosomal repeat sequences from Arabidopsis (Campellet al. 1992).
Nuclei isolation and run-on transcription assays: Nuclei were isolated from the kernel pericarps of ears sampled at 18 days after pollination (DAP). Pericarps were peeled from kernels and stored for up to 1 month at -20° in buffer containing 50% glycerol, 10 mm KCl, 10 mm MgCl2, 0.1 mm dithiothreitol, and 20 mm 2-[N-morpholino]ethane sulfonic acid, pH 6.0. For isolation of nuclei, ∼12-20 pericarps were removed from tissue storage buffer and gently blotted to remove excess buffer. Nuclei were prepared as previously described (Galbraithet al. 1983) by finely chopping the tissue in a plastic petri dish with a single-edged razor blade and filtering the cellular debris through 60- and 20-μm nylon filters. The nuclei were pelleted by centrifugation at 1000 rpm for 10 min at 4° (JS-4.3 swinging bucket rotor; Beckman Coulter, Fullerton, CA). After decanting the supernatant, the pellet was resuspended in nuclei storage buffer (50% glycerol, 5 mm MgCl2, 50 mm Tris-HCl, pH 8.5). Alternatively, nuclei were isolated as described by Dorweiler et al. (2000), on the basis of the modified chromatin isolation protocol of Steinmuller and Apel (1986). All isolated nuclei were stored at -80°. Run-on transcription reactions were performed as described by Cone et al. (1993), using 5 × 106 to 8 × 106 nuclei per reaction. The reactions were treated with DNaseI (RNase-free, 27 μg per reaction) and proteinase K and extracted with phenol:chloroform:isoamyl alcohol (100:100:1). The labeled RNA was precipitated by adding 1/10 vol of 3 m sodium acetate, pH 5.0, and 2 vol of 100% ethyl alcohol. The supernatant was discarded and the pellet was briefly air dried and dissolved in 100 μl of H2O. Samples were further purified by passage through a Micropure-EZ spin column (Millipore, Bedford, MA) and then through a Microcon 30 spin column (Millipore) according to the specifications of the manufacturers. The reaction products were hybridized to either DNA gel blots (Coneet al. 1986) or slot blots. For the DNA gel blots, 4 μg of plasmid DNA was digested to release an insert fragment. For slot blots, 100 ng of denatured purified gene fragments or the equivalent amount of linear plasmid vector DNA was added per slot and transferred to nitrocellulose membrane using a slot blotter. Plasmids used include: pWRP59, containing a 315-bp P1-wr cDNA (position +1080 to +1394) fragment inserted in pBluescript II SK(-) (Chopraet al. 1996); pC2, containing a maize c2 gene cDNA (Wienandet al. 1986); pA1, containing a maize a1 gene cDNA (Schwarz-Sommeret al. 1987); pChi, containing a maize chi1 gene cDNA (Grotewold and Peterson 1994); and pZMU14, containing a genomic clone of the maize ubiquitin gene (Christensenet al. 1992). Results were quantified using phosphorimager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA). Hybridization values for each gene were normalized by subtracting any background hybridization to plasmid DNA and dividing by the hybridization value of ubiquitin.
—Vegetative and floral organ pigmentation patterns of P1-wr Ufo1 (left, A-D) and P1-wr ufo1 (right, A-D). Mature ear (A), leaf blade and leaf sheath junction (B), tassel (C), silk and husk on an immature ear (D).
Mapping of Ufo1: Fifty-three backcross progeny from six families of (W23 Ufo1 × W23)BC5 were grown in a greenhouse, and the Ufo1 phenotype was scored visually on the basis of red pigmentation in the husk and kernel pericarp. DNA was extracted from lyophilized leaf tissue using the modified CTAB method. Methods and protocols of PCR amplification using simple sequence repeat (SSR) primers and DNA gel electrophoresis and blotting are described elsewhere (http://www.maizemap.org/resources.htm). The relative positions and linkage of polymorphic SSR markers were established using the three-point command of MAPMAKER/EXP, version 3.0b.
RESULTS
Ufo1 is an allele-specific modifier of p1 expression: Charles Burnham isolated maize plants with variable orange plant color; these were subsequently shown to contain a modifier of p1 expression that was named Unstable factor for orange1 (Ufo1; Styles 1982; Styleset al. 1987). Genetic tests show that Ufo1 is a dominant factor that segregates independently from p1. Plants that carry P1-wr and Ufo1 have pigmented pericarp and more intense cob color compared with plants of genotype P1-wr ufo1, which have colorless pericarp and red cob glumes (Figure 1). In the pericarp, pigmentation is often strongest in and around the silk attachment point on the kernel. Additionally, plants of P1-wr Ufo1 genotype commonly have intense pigmentation in other organs including dried silk, tassel glumes, husk, and leaf sheath (Figure 1). In addition to the gain of pigmentation, P1-wr Ufo1 plants exhibit pleiotropic effects in which plants are variably stunted and weak. The enhanced pigmentation effect of Ufo1 was observed in several inbred lines carrying P1-wr, including W22, W23, and B73. In contrast, Ufo1 had no effect on pigmentation conferred by other p1 alleles tested, including P1-rr, P1-rw, and P1-ww. A notable exception is P1-rr-CFS327, which shows variegated gain of pericarp pigmentation phenotype in the presence of Ufo1 (see below). On the basis of these and previous genetic studies (Styleset al. 1987), we hypothesized that Ufo1 increases the levels of P1-wr expression and also induces ectopic expression of P1-wr in floral and vegetative tissues.
Map position of Ufo1: The Ufo1 locus was genetically mapped in a population of 53 progeny derived from six families representing the fifth-generation backcross with inbred line W23. Progeny plants were scored for pigmentation phenotype and analyzed for genotype using SSR markers (materials and methods). Twenty-three of the 53 backcross progeny expressed the Ufo1 phenotype (Figure 2). Despite the possibility of variable expressivity and incomplete penetrance (see below), this ratio is not significantly different from the expected Mendelian segregation ratio of one locus determining Ufo1 effects (P = 0.55). However, the difference between recombination ratios of the Ufo and the wild-type phenotypic classes for umc1367 and umc1179 markers provides evidence that some of the putative recombinants in the wild-type phenotypic class might be due to incomplete penetrance. Previously the maize genome was surveyed for linkage to Ufo1, and linked loci were found along the short arm of chromosome 10 (Matzet al. 1991). Our data further narrow the location of the Ufo1 locus to bin 3 on chromosome 10. Three-point analysis placed the Ufo1 locus between umc1576 and umc1367 if all 53 individuals were included and coincident with umc1367 if only the definitive Ufo expressers are considered.
Ufo1 increases levels of P1-wr transcripts: Previous studies have shown that p1 directly regulates transcription of the maize a1 gene that is required for phlobaphene pigment biosynthesis (Grotewoldet al. 1994). To test whether Ufo1 acts by affecting p1 transcript levels, we compared the levels of p1 and a1 transcripts in plants carrying P1-wr Ufo1 with those with P1-wr, P1-ww, and P1-rr alleles in the absence of Ufo1. Steady-state transcript levels were determined by Northern analysis of poly(A)+ RNA isolated from developing kernel pericarp. The blot was sequentially hybridized to radioactively labeled DNA fragments from the maize p1, a1, and actin genes (Figure 3A). Hybridization signals were quantified by densitometry and normalized to actin transcript levels. In agreement with our previous results (Chopraet al. 1996), P1-wr pericarp has reduced levels of p1 and a1 transcripts as compared with P1-rr pericarp. Pericarps of P1-wr Ufo1 plants, however, have a threefold increase in p1 transcripts and a twofold increase in a1 transcripts relative to P1-wr pericarps (Figure 3B). Similar elevated transcript levels were also observed in other pigmented tissues of P1-wr Ufo1 plants, including cob glumes, developing husk, and silks (data not shown). Thus, Ufo1-induced pigmentation of pericarp and other tissues is associated with increased p1 and a1 transcript levels.
—Mapping of Ufo1 on chromosome 10 of maize. (A) Number of individuals in each genotype/phenotype class for four markers in the (W23Ufo1 × W23)BC5 population. Under the genotype column A denotes plants homozygous for W23 alleles and H indicates that the plants were heterozygous for W23 and W23Ufo1 alleles at that SSR locus. (B) Position of the Ufo1 locus when only the 31 Ufo-expressing plants are considered. Markers are indicated in italics, with the chromosome and bin position underneath. For arrangement of SSR markers relative to maize chromosome 10 consensus restriction fragment length polymorphism markers, see the Missouri Maize Project web site (http://www.maizemap.org/maps.htm).
To determine whether Ufo1 affects P1-wr transcription rate or transcript stability, we performed nuclear run-on transcription assays using nuclei isolated from kernel pericarp of P1-wr Ufo1, P1-wr, P1-ww, and P1-rr genotypes. The labeled nascent transcripts were detected by hybridization to cDNA sequences immobilized on nylon membranes (materials and methods). The genes tested included c2, chi, and a1, which encode enzymes for flavonoid pigment biosynthesis; a P1-wr cDNA fragment lacking the MYB domain to avoid cross-hybridization with other Myb genes; and a maize actin gene for normalization. The mean transcription rates determined in two experiments are shown in Figure 3C. The P1-wr and P1-rr alleles exhibit similar transcription rates, indicating that the reduced P1-wr steady-state transcript levels relative to P1-rr transcript levels (Figure 3A; Chopraet al. 1996) must be due to increased turnover of P1-wr transcripts relative to P1-rr transcripts; i.e., P1-wr transcripts are less stable than P1-rr transcripts. In the case of P1-wr Ufo1 pericarp, the transcription rate for P1-wr was on average threefold greater than that for P1-wr and P1-rr plants. Thus, presence of the Ufo1 factor increases P1-wr transcription in pericarp. However, because p1 steady-state transcript levels in P1-wr Ufo1 pericarp are similar to those of P1-rr pericarp, Ufo1 likely does not affect the rate of P1-wr message turnover. Thus, Ufo1 increases P1-wr transcripts, but does not affect P1-wr transcript stability; the net effect is that levels of P1-wr Ufo1 transcript levels and P1-rr transcript levels are approximately equal in kernel pericarp.
The increased level of P1-wr transcripts in P1-wr Ufo1 compared with P1-wr ufo1 is also associated with increased transcription rates of the c2, chi, and a1 genes. This observation is consistent with the previously reported role of p1 in activating transcription of these genes (Grotewoldet al. 1994). It is interesting that the rates of transcription of c2, chi, and a1 are actually greater in P1-wr Ufo1 than in P1-rr (Figure 3C), even though the steady-state levels of P1-wr transcripts in the presence of Ufo1 are approximately equal to that of P1-rr (Figure 3B). It seems unlikely that Ufo1 directly affects transcription of these genes independently of its action on P1-wr, because Ufo1 does not enhance pigmentation in the presence of P1-rr. One possible explanation is that hypertranscription of c2, chi, and a1 in the P1-wr Ufo1 genotype reflects a higher activating potential of the P1-wr protein relative to the P1-rr protein; the proteins differ significantly at their C termini due to structural differences in the two genes, although there is no other evidence to suggest that they differ in their transcriptional activation potentials (Chopraet al. 1996). More surprising, despite a higher rate of a1 transcription in the P1-wr Ufo1 genotype relative to P1-rr, the steady-state levels of a1 transcripts are similar in the P1-wr Ufo1 and P1-rr genotypes. Thus, it seems that while a1 transcription rate is increased in P1-wr Ufo1, the stability of a1 transcripts is decreased. While there are a number of possible explanations for this observation, it would be interesting to determine whether increased flux through the flavonoid biosynthetic pathway, as in the case of P1-wr Ufo1, triggers a negative feedback mechanism that destabilizes the transcripts encoding enzymes of the pathway. Such a model could be tested by analysis of P1-wr Ufo1 plants, which also carry a c2 mutation that would block the first committed step of the flavonoid pathway.
P1-wr is demethylated in the presence of Ufo1: In previous studies, we have shown that certain P1-wr coding sequences and flanking regions are resistant to cleavage by methylation-sensitive enzymes, while the corresponding sequences in P1-rr are readily cleaved (Chopraet al. 1998). These results indicate that many sequences in P1-wr are hypermethylated relative to P1-rr. Here, we tested whether the Ufo1-induced overexpression of P1-wr was associated with a change in P1-wr DNA methylation. Southern analysis of P1-wr ufo1 and P1-wr Ufo1 leaf genomic DNA digested with several restriction enzymes were hybridized to p1 probe fragments (see Figure 4 for position of p1 gene probe fragments). The hybridization patterns produced by p1 probe fragment 15 are shown in Figure 4A. The absence of any restriction polymorphism between P1-wr ufo1 and P1-wr Ufo1 DNA digested with EcoRI, HindIII, KpnI, and SacI indicates that no changes in the gross P1-wr gene structure occurred in the presence of Ufo1. To test for alterations in DNA methylation, leaf DNA samples from P1-wr ufo1 or P1-wr Ufo1 plants were digested with the restriction enzyme isoschizomer pair MspI and HpaII, differing in their sensitivity to CG methylation. Southern hybridizations were performed using p1 fragments that detect the entire 12.6 kb of each of the six P1-wr gene copies. The results of hybridization with three such fragments are shown in Figure 4B. The p1 probe fragment 15 gives a similar hybridization pattern to MspI digests of both P1-wr Ufo1 and P1-wr ufo1 DNA, with the exception of a 500-bp fragment, which is prominent in the Ufo1 sample but absent from the ufo1 DNA. HpaII digestion, however, produces approximately seven fragments in P1-wr Ufo1 DNA, which are absent in the P1-wr ufo1 sample. The p1 probe F-6 detects fragments of ∼4.5 and 0.6 kb that are present in HpaII digests of P1-wr Ufo1, but absent from P1-wr ufo1. Similarly, probe F-8B, which is part of the second intron of the P1-wr gene, detects fragments of 2.6 and 1.2 kb in HpaII-digested P1-wr Ufo1 DNA, whereas these fragments are absent or present at much-reduced levels in P1-wr ufo1 DNA. Additionally, probe fragment F-13, which is derived from the third exon of p1, detects four fragments ranging in size from 1900 to 3600 bp in HpaII-digested P1-wr Ufo1 DNA, whereas probe F-13 detects a fragment of 8.1 kbp in HpaII-digested P1-wr ufo1 DNA (not shown). These results are summarized in Figure 5, which shows a comparison of the CpG methylation status of the P1-wr HpaII sites in the presence or absence of Ufo1. These results indicate that most of the HpaII sites in P1-wr ufo1 are methylated and resistant to HpaII digestion. In the presence of Ufo1, the P1-wr DNA is more sensitive to HpaII digestion and therefore less methylated: some sites become fully demethylated, while other sites undergo partial demethylation. In contrast, we detected one site (Figure 5D, site 1) that shows partially increased methylation in the presence of Ufo1. Still other sites are unchanged in the presence or absence of Ufo1, remaining as methylated, demethylated, or partially methylated. The changes in methylation patterns occurred in both coding and noncoding regions of P1-wr. Due to the multicopy nature of the P1-wr allele, sites showing partial demethylation may reflect heterogeneity within the individual gene copies of the locus or among the different cells used to prepare DNA. Interestingly, sites that show complete sensitivity or resistance to HpaII digestion must have the same methylation status among all of the six P1-wr copies; i.e., they exhibit coordinate methylation throughout the P1-wr complex. In summary, the presence of Ufo1 results in dramatic changes in the P1-wr methylation pattern, with most sites tested showing decreased methylation in the presence of Ufo1.
—Steady-state and nuclear run-on transcription analysis. (A) Northern analysis of p1, a1, and actin RNA levels from 18 DAP pericarp of P1-wr Ufo1, P1-wr, P1-ww, and P1-rr genotypes. Probes are indicated at the left. (B) Quantified RNA levels of p1 and a1 from hybridizations in A and normalized to actin to calculate relative transcript levels (y-axis). (C) Nuclear run-on assays. In vitro transcription data comparing the relative transcription rates of p1, c2, chi, and a1 genes in 18 DAP pericarp from P1-wr Ufo1, P1-wr, P1-ww, and P1-rr genotypes. Results were quantified using a phosphorimager. Error bars represent the standard error values based on two independent experiments.
Ufo1-induced hypomethylation and overexpression of P1-wr are correlated: Owing to the incomplete penetrance of the Ufo1 factor, we tested whether P1-wr hypomethylation and enhanced pigmentation were completely correlated. Prior to this analysis, the original Ufo1 stock was backcrossed four to five times with inbred lines W22 or W23 (both are P1-wr); this series of backcrosses was done to introgress Ufo1 into a known genotype and thereby remove the possible influence of other genetic factors that may have contributed to the phenotypic variation in Ufo1 expression. We then selected five parent plants with strong Ufo1-induced pigmentation in kernel pericarp and husk tissues; the ears of these plants were crossed with a P1-wr line, and their progeny were classified for Ufo expression on the basis of pigmentation of leaf sheath, husk, tassel glumes, pericarp, and cob glumes. Plants that showed pigmentation in tissues not normally pigmented in the recurrent inbred parent were classified as Ufo expressing, while plants that did not show any ectopic pigmentation were classified as nonexpressing (Table 1). In all progeny, the number of nonexpressing plants exceeded the number of expressing plants. Chi-square analysis showed that in two out of five backcross progeny, the ratio of Ufo-expressing vs. nonexpressing plants does not fit a 1:1 Mendelian ratio expected for a single dominant factor. In three out of the five families, the chi-square analysis does not cause rejection of a 1:1 ratio, but it is possible that testing of more individuals would likewise cause rejection of a 1:1 ratio. These results support the conclusion from the mapping data (see above) that Ufo1 is incompletely penetrant.
—Restriction mapping and DNA methylation analysis of P1-wr and P1-wr Ufo1 plants. (A) Southern hybridization of genomic DNA isolated from two (1, 2) independent P1-wr and P1-wr Ufo1 plants. The blot was hybridized to p1 genomic DNA fragment 15 (Lecheltet al. 1989). Restriction enzymes are shown at left and hybridizing band sizes are indicated in kilobases at right. (B) Southern hybridization of genomic DNA digests of P1-wr and P1-wr Ufo1 plants with HpaII (H) and MspI (M) restriction enzymes. Same blot was stripped and used to hybridize different DNA fragments of p1 gene as probes. Results are shown for fragments 15, 6, and 8B. For position of each probe fragment see Figure 4. Molecular weight marker fragments (in kilobases) are shown at left.
—DNA methylation of the P1-wr allele in Ufo1 and ufo1 plants. Cytosine methylation map deduced from Southern hybridization using MspI and HpaII digests of P1-wr and P1-wr Ufo1 genomic DNA. (A) The structure of the 12.6-kb tandem repeats of P1-wr. (B) A full 12.6-kb repeat and a partial distal repeat are shown. (C) Intron/exon organization. Exons are depicted as solid boxes. Open boxes represent untranslated regions. A bent arrow in the methylation map diagram represents transcription start site. (D and E) Methylation maps of P1-wr and P1-wr Ufo1 plant DNA. Numbers above the methylation map represent specific CpG sites discussed in the text. Restriction fragments generated by the HpaII digest and probes (numbered boxes) used to map them are shown below the methylation map of P1-wr. Pattern of methylation states of HpaII sites are shown as solid, open, and hatched circles.
A subset of plants from each backcross progeny was further tested for Ufo expression and P1-wr methylation status (Table 1). Genomic DNA was isolated from seedling leaves, digested with HpaII, and used for DNA gel blots. The DNA gel blots were hybridized with p1 probe F-15. The hybridization patterns were used to infer the cytosine methylation status of P1-wr in each plant. The seedling methylation status was then compared with the pigmentation state of each plant scored at maturity. The results show that the P1-wr methylation state and Ufo expression phenotype are completely correlated: all the expressing plants show hypomethylation, while the nonexpressing plants show hypermethylation of the P1-wr sequences. None of the plants belonged to the expressing and hypermethylated class or the nonexpressing and hypomethylated class. Hybridization results obtained from 17 progeny plants of backcross family 2 (Table 1) are presented in Figure 6. Lanes 1 and 2 contain DNA from P1-ww Ufo1/- and P1-wr ufo1 plants, respectively. Eight plants with normal P1-wr pigmentation (R) had hybridization patterns similar to that of standard P1-wr (lane 2). In contrast, nine plants with enhanced pigmentation (U) showed enhanced sensitivity to digestions with HpaII. The mature ears produced by the progeny plants showed considerable variation in pigmentation, ranging from darkly pigmented pericarp and cob (plant 12) to moderately variegated pericarp (plants 4 and 6) to a near-P1-wr-like pattern (plant 3). However, close examination shows that plants with a near-P1-wr-like pattern have small red sectors on kernels and husks (Figure 6, arrow lower left). Interestingly, the DNA gel blot hybridization patterns of such plants resemble that of the standard P1-wr allele except for the presence of a 503-bp fragment (Figure 6, lane 3). The 503-bp fragment is produced by digesting at hypomethylated HpaII sites 2 and 3, located ∼5 kbp 5′ of the p1 transcription start site (Figure 5). Overall these results confirm that Ufo1-induced pigmentation is highly correlated with reduced methylation of HpaII sites in the P1-wr gene complex.
Instability of Ufo1 and its effects on P1-wr: To further investigate the inheritance and stability of the effects of Ufo1 on P1-wr expression, we analyzed the selfed and outcross progeny of the four plants whose ears are pictured in Figure 6. Each of the four plants was self-pollinated and outcrossed to a standard inbred P1-wr line. Twenty seeds from each of the four self-pollinated ears and the corresponding outcrosses were grown to maturity. Genomic DNA was tested for methylation by HpaII digestion, and plants were scored for ectopic pigmentation. Ear 3 (Figure 6) produced progeny plants, all of which were of standard P1-wr methylation and pigmentation pattern, indicating that the effect of Ufo1 observed in the previous generations was now undetectable. Additionally, there were no Ufo-expressing plants in the progeny of the outcross to a standard P1-wr line [W23]. The P1-wr stock used for outcross with the Ufo1 parent was naïve, i.e., it had not previously been exposed to Ufo1, and hence it could not have become refractory to activation by Ufo1. We conclude that the incomplete penetrance of the Ufo1-induced activation of P1-wr results from loss of Ufo1 function.
The self-pollinated ears 4, 6, and 12 produced P1-wr Ufo1 and P1-wr ufo1 plants in the ratio of 4:16, 6:14, and 6:14, respectively. These numbers do not fit the 3:1 ratio expected from segregation of a single dominant factor. The corresponding outcross progenies confirmed the presence of the Ufo1 factor in these plants, although the numbers of Ufo1 to ufo1 plants again differed from a 1:1 ratio (data not shown). Many of the progeny ears produced by both self-pollination and outcross showed variegated or less extensive pigmentation compared with their parental ears. Similar to the results shown in Figure 6, the degree of hypomethylation observed by DNA gel blot analysis was strongly correlated with the intensity of pigmentation of mature plant tissues. Overall, these results indicate that Ufo1 is highly unstable and spontaneously changes to a state that does not activate P1-wr. It is unknown whether Ufo1 can become reactivated following its loss of function.
It is uncertain whether Ufo1 can be maintained in a homozygous condition. Due to the high level of spontaneous inactivation, it has not been possible to demonstrate by progeny analysis that any individual plant is a Ufo1 homozygote. However, we did not observe 25% kernel abortion or 25% seedling lethality in the progeny of self-pollinated Ufo1 plants as would be expected if Ufo1 were homozygous lethal. Some Ufo-expressing plants are severely stunted and died before maturity; whether these highly affected plants represent the Ufo1 homozygous class could be determined by molecular analysis for inheritance of a linked marker, although this analysis has not yet been done.
Ufo1 does not induce genome-wide demethylation: To test whether Ufo1 may affect methylation of other genomic sequences, DNA gel blots prepared from HpaII- and MspI-digested P1-wr Ufo1 and P1-wr ufo1 leaf DNA were hybridized to three probes that detect repetitive maize sequences: a 185-bp repeat sequence from a maize chromosomal knob; a 5S rDNA from Arabidopsis; and a fragment of maize subtelomeric sequence, which also cross-hybridizes with a repetitive sequence present near the maize p1 gene on chromosome 1S. No detectable difference in methylation of these repeat sequences was observed between P1-wr and P1-wr Ufo1 plants (Figure 7). We conclude that Ufo1 does not induce genome-wide demethylation of repetitive sequences.
P1-wr Ufo1 × P1-wr backcross progeny analysis for Ufo expression and P1-wr DNA methylation
—Association of P1-wr methylation pattern and gain of pigmentation phenotype in the presence of Ufo1. Southern analysis of HpaII-digested leaf genomic DNA of sibling progeny plants from a backcross (P1-wr/P1-wr Ufo1/-× P1-wr/P-wr) was hybridized to probe fragment 15. Standard DNA samples included are P1-ww Ufo1 (W) and P1-wr (R). Ear phenotypes were scored as U and R for a Ufo expresser and for nonexpresser plants, respectively. Ear phenotypes of four sibling plants and their corresponding DNA lanes are indicated. Molecular weight marker fragments in kilobases are shown at left. Arrowheads with sizes in base pairs indicate major hypomethylated bands hybridizing in P1-wr Ufo1 plant DNA.
DISCUSSION
We describe here a dominant factor named Ufo1 (Styleset al. 1987), which modifies the organ-specific expression patterns of the P1-wr allele. This modifier was originally identified because of its ability to induce orange-red pigmentation in vegetative and floral tissues of maize plants (Styles 1982). Our results confirm the finding by Styles et al. (1987) that Ufo1 alters pigmentation only in inbred lines that carry a specific allele of the p1 gene, P1-wr. The P1-wr allele normally conditions a colorless kernel pericarp and red cob phenotype that is very common among United States Corn Belt Dent varieties (Goodman and Brown 1988). A number of red-cobbed lines, including B37, B73, Oh43A, and W22, contain a multicopy P1-wr allele indistinguishable from the type found in W23 (S. Chopra, M. McMullen and T. Peterson, unpublished data). In addition to conferring cob pigmentation, P1-wr has been identified as a major quantitative trait locus (QTL) controlling levels of silk maysin, a C-glycosyl flavone compound with antibiosis activity against corn earworm larvae (Byrneet al. 1996). Moreover, selection for grain yield in a population segregating multiple p1 alleles resulted in significant increases in P1-wr allelic frequency (Frascaroli and Landi 1998). Thus, the P1-wr allele is widespread, beneficial, and largely stable in its expression pattern. In contrast, the presence of Ufo1 induces dramatic and variable alterations in P1-wr expression. For example, within a single P1-wr Ufo/-family, pericarp pigmentation can range from deep red through various degrees of variegated red to colorless. Husks are commonly variegated with prominent red and white sectors; variegation can also be observed on leaf sheath and tassel branches. Moreover, while the P1-wr expression pattern is stable in diverse genetic backgrounds, the Ufo1-induced pigmentation varies markedly in intensity and uniformity in different genetic backgrounds. These observations suggest that additional genetic factors influence the interaction of Ufo1 and P1-wr.
—Ufo1 does not affect global DNA methylation. Gel blots prepared from HpaII (H) and MspI (M) digested P1-wr Ufo1 and P1-wr leaf DNA were hybridized to repetitive probe DNA fragments: 5S rDNA from Arabidopsis (5S), maize 185-bp repeat sequence (185), and maize MTY7SC (mty7) repeat sequence.
Ufo1 induces P1-wr transcription: The P1-wr locus contains six highly similar gene copies in a tandem direct array (Chopraet al. 1998), whereas the steady-state level of P1-wr transcripts in developing pericarp tissue is only ∼30% of the level present in P1-rr pericarp (Figure 3; Chopraet al. 1996). If each of the six gene copies of P1-wr were equally expressed, then expression of each copy would be reduced to ∼5% of the level of the single-copy P1-rr gene. However, run-on transcription analysis determined that the numbers of nascent p1 transcripts are similar for the P1-rr and P1-wr alleles (Figure 3C). This suggests that the lower steady-state level of P1-wr transcripts is due to increased RNA turnover. In the presence of Ufo1, the transcription rate of P1-wr increases approximately threefold, and the steady-state levels of P1-wr transcripts also increase approximately threefold to a level approaching that of P1-rr pericarp. These results indicate that Ufo1 increases the number of nascent transcripts of the P1-wr allele. Because P1-wr is multicopy, this increase could come about by increasing transcription among multiple copies or by activating a subset of silenced copies. In addition, these results demonstrate that upregulation of P1-wr RNA is sufficient to bring about a correlative gain of red pericarp pigmentation and enhanced cob color. This indicates that transcription of the structural genes required for phlobaphene pigmentation is limited by the level of p1 expression. This conclusion is supported by previous observations that P1-wr homozygous plants have darker red cob color than plants in which P1-wr is heterozygous with a null p1 allele (Brink 1958; Athma and Peterson 1991). Moreover, recent QTL studies have found that the p1 locus has an additive effect on control of silk maysin levels (Byrneet al. 1996).
Variegated pigmentation and DNA demethylation in P1-wr Ufo1 plants: Our genetic and molecular data show that the presence of Ufo1-induced pigmentation is strongly correlated with P1-wr overexpression and reduced methylation of P1-wr DNA. Each of the six P1-wr gene copies has the same pattern of methylation (Chopraet al. 1998). The P1-rr allele, which does not show any obvious interaction with Ufo1, is single copy and considerably less methylated than the P1-wr allele (Chopraet al. 1998). Further, we observed that the degree of enhanced pigmentation phenotype is strongly correlated with the degree of P1-wr demethylation. For example, a slight gain of P1-wr function indicated by a few red-striped kernels (Figure 6, ear 3) was associated with partial demethylation at two sites in P1-wr to release a 503-bp fragment upon digestion with HpaII. Plants with strong pericarp pigmentation, however, exhibited a much greater extent of demethylation of P1-wr sequences (Figures 5 and 6). Previous studies have also identified a strong correlation of p1 methylation and gene inactivation. In one study, an epiallele of P1-rr termed P1-pr exhibited suppressed pericarp pigmentation and reduced p1 transcription. The P1-pr epiallele was associated with hypermethylation of sequences within a 1.2-kb enhancer fragment located 5 kb upstream of the transcription start site (Das and Messing 1994). A DNAseI-sensitivity assay showed that changes in chromatin conformation occurred within the 1.2-kb distal enhancer fragment of the suppressed (P1-pr) vs. active (P1-rr) allele of p1 (Lundet al. 1995). In P1-wr, the sequences that would correspond to the 1.2-kb enhancer of P1-rr are truncated; however, two HpaII sites border this truncated enhancer region in P1-wr (Figure 5, sites 2 and 3; Chopraet al. 1998). These two CpG sites are methylated in standard P1-wr, but they are demethylated in both the strong and the minimal Ufo1 individuals as evidenced by the appearance of the 503-bp HpaII fragment (Figure 6, ear 3). Further investigation will be required to determine whether these sequences represent a critical p1 regulatory region.
It is unclear whether the observed demethylation of P1-wr sequences is a direct effect of Ufo1 or a secondary effect that results from activation of P1-wr transcription. A potential role of DNA methylation in gene silencing and HDGS has been demonstrated through the isolation and characterization of the ddm1 and hog1 mutations in Arabidopsis (Furneret al. 1998; Kakutaniet al. 1999). The DDM1 gene encodes a protein with homology to SWI2/SNF2-like proteins, and it has been suggested that it may affect DNA methylation by regulating chromatin structure (Jeddelohet al. 1999). Unlike the ddm1 mutation, our results indicate that Ufo1 does not affect methylation of repetitive sequences other than P1-wr (Figure 7). Interestingly, recent analysis of the mom1 mutation in Arabidopsis demonstrates the existence of an alternative type of epigenetic regulation that mediates transcriptional silencing independently of changes in DNA methylation (Amedeoet al. 2000).
Allele specificity of the p1-Ufo1 interaction: A striking feature of Ufo1 is its allele-specific effects on pigmentation. As noted above, Ufo1-induced pigmentation was observed only with P1-wr alleles, with the exception of a novel allele termed P1-rrCFS327. This latter allele normally specifies grainy red pericarp and red cob. However, P1-rrCFS327 gives distinct pericarp, husk, and vegetative tissue pigmentation in the presence of Ufo1. Interestingly, P1-rrCFS327 has a multicopy gene structure resembling that of standard P1-wr, but which is hypomethylated relative to standard P1-wr. In other words, P1-rrCFS327 appears to be a semistable epiallele of P1-wr with reduced methylation and increased expression of P1-wr (Coccioloneet al. 2001). Thus, Ufo1 may affect only p1 alleles that have a multicopy structure. Alternatively, the allele specificity of Ufo1 could reflect sequence differences in the P1-rr and P1-wr alleles located >5 kb upstream and 8 kb downstream of the transcription start site.
Allele-specific differences were also observed in a study of p1 paramutation: P1-rr expression is suppressed by exposure to a transgene locus that carries a 1.2-kb enhancer fragment derived from the P1-rr allele. Moreover, the suppressed state is associated with P1-rr hypermethylation (Sidorenko and Peterson 2001). In contrast, P1-wr is not affected by exposure to the paramutagenic locus, nor does P1-wr transmit a paramutagenic signal (Sidorenko and Peterson 2001). Thus, the P1-rr and P1-wr alleles exhibit distinct and complementary susceptibilities to Ufo1 activation and transgene-induced paramutation.
Pleiotropism and possible function of Ufo1: As mentioned above, P1-wr Ufo1 plants exhibit variable defects in growth and vigor. The degree of stunted growth is proportional to the intensity of plant pigmentation, leading to an earlier suggestion that production of phlobaphene pigments in vegetative tissues where they do not normally accumulate may be deleterious to the plants (Styleset al. 1987). This idea is supported by the observation that transgenic maize plants that express p1 transgenes in vegetative tissues show similar deleterious effects (Coccioloneet al. 2001; S. M. Cocciolone, unpublished data). In addition to its effects on P1-wr, Ufo1 could conceivably affect other genes whose altered expression may lead to other pleiotropic effects. Pleiotropic developmental abnormalities have also been observed in maize plants carrying mutations in mediator of paramutation1 (mop1), which is involved in establishment and maintenance of paramutation at several maize loci required for anthocyanin biosynthesis (Dorweileret al. 2000). The reported pleiotropic effects of mop1 include short stature, delayed flowering, and tassel feminization, whereas Ufo1 seems to affect primarily plant stature. Like Ufo1, the mop1 mutation does not elicit global genome demethylation (Dorweileret al. 2000), but it does lead to demethylation of silenced Mutator transposon sequences (Lischet al. 2002). It will be interesting to determine whether Ufo1 may similarly affect Mutator transposon methylation.
We considered the possibility that ufo1 encodes or controls a trans-acting factor whose normal function is to activate p1 expression in floral organs. In this model, the Ufo1 allele would represent a hypermorph that is overexpressed in floral organs and ectopically expressed in vegetative tissues. This model seems unlikely due to the allele specificity of Ufo1 action. The P1-rr and P1-wr coding and regulatory sequences are highly similar: in the promoter region, P1-rr and P1-wr are 99% identical over a 5-kb region 5′ of the transcription start sites. Thus, it seems likely that P1-rr and P1-wr would be activated by the same regulatory factors. Yet, Ufo1 has no detectable effect on expression of a standard P1-rr allele, even though enhancement of P1-rr expression could easily have been observed in husks and vegetative tissues of P1-rr plants.
The results of this and previous studies indicate that transcription of the P1-wr gene complex is suppressed in kernel pericarp and that this suppressed state is associated with P1-wr hypermethylation. We propose that Ufo1 alleviates the transcriptional suppression of P1-wr; the associated demethylation of P1-wr sequences may be a secondary effect of transcriptional activation. Ufo1 may encode or regulate a factor that modifies the chromatin structure of P1-wr and possibly other multicopy complex loci. Genes that putatively affect chromatin structure have been mapped recently in maize (http://www.chromdb.org). Two of these genes map in the vicinity of Ufo1 in bin 10.03, chromosome 10: sdg108b is related to SET domain genes, some of which encode histone methyltransferases (Reaet al. 2000), and chr109a is related to chromatin-remodeling complex subunit R (SWI2/SNF2; http://www.chromdb.org). Due to the highly unstable nature of Ufo1, it would be impractical to use transposon tagging for its molecular isolation; hence it will be interesting to determine whether Ufo1 represents a mutant allele of one of these candidate genes. Additionally, it will be important to test whether Ufo1 can activate the expression of other maize genes, both natural and transgenic, that contain multicopy repeat sequences.
Acknowledgments
We thank Derek Styles and Benjamin Burr for providing us the stocks of Ufo1 and Maize Genetics Cooperation Stock Center (Urbana, IL) and National Seed Storage Laboratory for the maize germplasm. We thank Terry Olson for excellent technical assistance. Research for the characterization of the Ufo1 allele was supported by a National Science Foundation grant to T.P. and was partly supported under the Hatch project no. 3855 research funding to S.C. from the Department of Crop & Soil Sciences and Life Sciences Consortium of the Pennsylvania State University, and research funding to M.D.M. from USDA-ARS.
Footnotes
-
Communicating editor: V. Sundaresan
- Received June 24, 2002.
- Accepted December 18, 2002.
- Copyright © 2003 by the Genetics Society of America
