RNA-Dependent RNA Polymerase Is Required for Enhancer-Mediated Transcriptional Silencing Associated With Paramutation at the Maize p1 Gene
Lyudmila Sidorenko, Vicki Chandler


Paramutation is the ability of an endogenous gene or a transgene to heritably silence another closely related allele or gene. At the maize p1 (pericarp color1) gene, paramutation is associated with decreases in transcript levels and reduced pigmentation of the endogenous allele that normally specifies red seed coat (pericarp) and cob pigmentation. Herein we demonstrate that this silencing occurs at the transcriptional level and that a specific enhancer fragment from p1 is sufficient to induce all aspects of paramutation. Further, we demonstrate that a mutation in the mop1 gene (mediator of paramutation1), which encodes a RNA-dependent RNA polymerase, is absolutely required for establishing the silencing associated with p1 paramutation. In contrast to its effects on other paramutation loci, the mop1 mutation does not immediately reactivate a previously silenced allele; several generations in the presence of the mop1 mutation are required. In addition, the mop1 mutation was also able to release tissue-specific silencing of another p1 allele that does not participate in paramutation, but does contain a tandem repeated structure and is likely regulated through epigenetic mechanisms. These results demonstrate that RNA-mediated gene-silencing mechanisms play key roles in p1 paramutation and the spectrum of roles for MOP1 is broadened to include tissue-specific expression patterns.

THE term “paramutation” was coined by Alexander Brink (1956) to describe a heritable silencing phenomenon that he observed in crosses with specific r1 (red1) alleles specifying dark and light anthocyanin kernel pigmentation. Subsequent to Brink's initial report, paramutation has been described for three other maize genes involved in the flavonoid biosynthetic pathway (Coe 1959; Hollick et al. 1995; Sidorenko and Peterson 2001). Two of these genes, b1 (booster1) and pl1 (plant color1), regulate biosynthesis of purple anthocyanins, while the third gene, p1 (pericarp color1), controls accumulation of red phlobaphene pigments. Related phenomenona involving endogenous genes and transgenes have been reported in multiple species, including other plants, fungi, and animals (Chandler 2007; Matzke et al. 2007; Zaratiegui et al. 2007). This study focuses on paramutation at the maize p1 gene, which encodes a myb-like transcriptional factor that activates a subset of biosynthetic genes responsible for production of red phlobaphene pigment in floral tissues of maize plants (Grotewold et al. 1994). Phlobaphene pigmentation is most pronounced in the mature ear in maternal tissues such as the pericarp (the seed coat covering the maize kernel) and the cob glumes (modified flower bracts embracing each kernel). Differential pigmentation of the pericarp and cob glumes distinguishes many naturally occurring p1 alleles, which are classified using two-letter suffixes, with the first letter denoting pericarp and the second letter signifying cob pigmentation. For example, an allele with a white pericarp and a red cob is denoted as P1-wr, while an allele with a white pericarp and a white cob is denoted as P1-ww (Figure 1A). The p1 allele that undergoes paramutation has a red pericarp and a red cob and is called P1-rr (Figure 1A), while silenced epialleles resulting from paramutation are designated P1-rr′ (Figure 1B).

Figure 1.—

Phenotypes of the endogenous p1 alleles and structure of P1-rr. (A) The phenotypes of endogenous alleles showing cob and pericarp pigmentation. (B) Silencing phenotypes of P1-rr′, ranging from orange (weakly silenced) to colorless (completely silenced). (C) The P1-rr coding sequence is flanked by long 5.2-kb direct repeats (solid rectangles) that overlap with smaller 1.2-kb direct repeats (hatched boxes). The 1.6-kb transposon interrupting the upstream-most 1.2-kb repeat is shown as an open triangle. Structure of P1-rr mRNA is shown above the locus map with shaded boxes denoting exons 1–3; the open box indicates alternatively spliced exon 4, and the solid line represents introns. Downstream copies of the P1.2 fragment overlaps with P1-rr exons 3 and 4. The bent arrow indicates the P1-rr transcription start site. The SalI restriction sites within 1.2-kb direct repeats delimit the long-distance enhancer fragment (P1.2). Positions of the P1.2 fragment, as well as the proximal fragments (P1.0 and P2.0, which also have enhancer activities, but show no paramutation activity as transgenes) and the basal P1-rr promoter and 5′-UTR (Pb), are indicated below the map.

Paramutation at the p1 gene was first described in crosses that brought together the endogenous P1-rr allele and a transgene, designated P1.2b∷GUS, which carried the P1.2 enhancer fragment, the basal promoter, and the 5′ untranslated leader of P1-rr (regions are diagrammed in Figure 1C) fused to a GUS reporter gene (Sidorenko and Peterson 2001). In these crosses, a subset of transgenic plants displayed patterned pigment or colorless pericarps (P1-rr′ in Figure 1B) while all nontransgenic siblings remained uniformly red. In subsequent generations, in the absence of an inducing transgene, silencing was highly heritable and secondarily paramutagenic; i.e., newly silenced P1-rr′ alleles were capable of silencing a naive P1-rr allele.

P1-rr has a complex structure with a single coding sequence flanked by two complex 5.2-kb direct repeats (Figure 1C). Each of the two 5.2-kb direct repeats overlaps with two 1.2-kb direct repeats, with the 5′-most 1.2-kb repeat truncated and interrupted by a 1.6-kb transposon-like sequence. Functional studies in transgenic plants determined that all three fragments (P1.2, P2.0, and P1.0, Figure 1C) harbored enhancer activity (Sidorenko et al. 2000). However, transgenes carrying fragments P1.0 and P2.0 (diagrammed in Figure 1C) of the P1-rr regulatory region did not induce silencing, indicating that sequences required for p1 paramutation are located within the P1.2 enhancer fragment (Sidorenko and Peterson 2001). The downstream 1.2-kb repeats partially overlap with the P1-rr transcript contributing the 3′-UTR to both the major transcript and an alternatively spliced transcript (Grotewold et al. 1991) (Figure 1C). The overlap between P1-rr transcripts and regulatory sequences creates a possibility for post-transcriptional and transcriptional gene silencing to play a role in paramutation. Molecular analysis of the P1.2b∷GUS-induced P1-rr′ silencing demonstrated a correlation with decreased transcript levels (Sidorenko and Peterson 2001). However, it remained unknown whether this decrease occurs at the transcriptional or post-transcriptional level. While the transgenes tested previously demonstrated that the P1.2 sequences were required for paramutation, the P1.2b∷GUS transgene also contained other p1 sequences such as the Pb fragment with the basal (235 bp) promoter and 5′-UTR (326 bp) of P1-rr that could also be contributing. Thus, experiments were performed to test whether P1-rr′ is silenced at the transcriptional level and if the P1.2 enhancer fragment is sufficient to induce paramutation.

The mop1 gene (mediator of paramutation1) is essential for paramutation at the b1, pl1, and r1 loci (Dorweiler et al. 2000) and is involved in the maintenance of Mutator transposon silencing and the silencing of certain transgenes (Lisch et al. 2002; McGinnis et al. 2006). The mop1 gene encodes a putative RNA-dependent RNA polymerase (RDR) most similar to Arabidopsis RDR2 (Alleman et al. 2006). RDR2 plays a critical role in RNA-directed DNA methylation and repressive chromatin modifications on certain transgenes, endogenous genes, and centromeric repeats that correlate with the production of 24-nt interfering RNAs (reviewed in Matzke et al. 2007; Zaratiegui et al. 2007). To determine if p1 paramutation involves similar mechanisms, experiments were carried out to determine whether p1 paramutation requires mop1.


Nuclear run-on analysis:

Nuclear run-on analysis was conducted using pericarps harvested 18 days after pollination. Pericarps were isolated from the upper half of an ear with the lower half left on the plant to develop mature seed pigmentation to confirm phenotype. Ten grams of pericarps were peeled into 10 ml of ice-cold extraction buffer [20 mm Tris–HCl, pH 7.8, 250 mm sucrose, 5 mm MgCl2, 5 mm KCl, 16% glycerol, 0.25% Triton X, 0.5 mm EGTA, 5 mm EDTA with 0.1 mm PMSF, and 0.1% β-mercaptoethanol (BME)] and chopped with a razor blade to a consistency of applesauce. Additional extraction buffer (30 ml) was added and the mixture was stirred for 30 sec to release nuclei, which were filtered through a 53-μm nylon mesh, followed by a rinse with ∼30 ml of ice-cold extraction buffer. The filtrate was centrifuged for 15 min at 6000 rpm, and pelleted nuclei were washed in 20 ml of ice-cold wash buffer (10 mm PIPES, pH 7.0, 10 mm MgCl2, 10% glycerol, and freshly added 0.1% BME) (Lisch et al. 2002; Carey et al. 2004) and resuspended in 2 ml of nuclei resuspension buffer (Hollick and Gordon 1993). The rest of the procedure was as described by Dorweiler et al. (2000). Nitrocellulose filters were prepared with 100 ng each of denatured DNA fragments prepared by PCR from plasmids. Primers and plasmid are available upon request. The sequences used were as follows: maize Ubiquitin2 (Christensen et al. 1992), a1 (AY105150), c2 gene (AY109395), and p1 gene (M73028). λ-Phage (100 ng) genomic DNA was used as negative control. Hybridization was as described by Dorweiler et al. (2000).

Plasmid constructions:

Previous studies used the P1.2b∷GUS transgene that carried the paramutagenic P1.2 fragment in a reverse orientation relative to the Pb fragment and the GUS gene (Sidorenko et al. 1999). We recloned P1.2 in a forward orientation to assure an equivalent comparison with other constructs used in this study. Constructs in which the Pb promoter was replaced with the 35S promoter were assembled in the intermediate plasmid pLP1001 that was derived by insertion of a synthetic linker in the PstI–NcoI-digested P0.4b∷GUS plasmid. The synthetic linker was created by annealing two oligos in equimolar (100 pmol/μl) concentration (gatgcggcgcgccatgcgtcgacatcgggatccatgcc and catgggcatggatcccgatgtcgacgcatggcgcgccgcatctgca) and contained PstI, AscI, SalI, BamHI, and NcoI restriction sites. The PCR fragment containing the 35S promoter and the Ω leader from cauliflower mosaic virus and maize Adh1 intron 1 was amplified from the pMCG161 plasmid (McGinnis et al. 2005). The −440 35S-containing fragment was amplified with primers carrying NcoI and SalI restriction enzyme tails: L26 (acgcgtcgaccatggtggagcacgacacttgg with SalI tail) and L28 (catgccatggccgcagctgcacgggtccaggaaagc with NcoI tail), while the −90 35S-containing product was obtained with L28 and L27 (catgccatggccgcagctgcacgggtccaggaaagc with SalI tail) primers. NcoI- and SalI-digested PCR products were inserted in pLP1001 that was opened with NcoI–SalI digestion. The pLP1144 and pLP1146 plasmids were digested with SalI and the P1.2 enhancer fragment was inserted in forward orientation to produce P1.2-440 35S and P1.2-90 35S, correspondingly. To make a promoterless P1.2∷GUS construct, P1.2-440 35S was digested with SwaI and BamHI to remove the −440 35S promoter, blunt ended with T4 DNA polymerase, and self-ligated. Modified fragments of the resulting constructs were verified by sequencing.

Plant transformation:

Transgenic plants were produced at the Iowa State Transformation Facility via biolistic bombardment (Frame et al. 2000). Plasmid DNA of interest was co-bombarded with selectable plasmid pBAR184(−), (Frame et al. 2000). Transgenic plants were produced in HiII germplasm (Armstrong 1994), which carries the null P1-ww allele, which is colorless. This allele carries a single copy of the P1.2 sequence (L. Sidorenko, unpublished data) and does not participate in p1 paramutation (L. Sidorenko, unpublished data). Transgenic callus clones resistant to the herbicide BASTA were screened by Southern blot hybridization to verify the presence of the intact P∷GUS expression cassette and to estimate the copy number of the transgene. Only clones containing at least one full-length P∷GUS transgene expression unit were regenerated.

Genetic stocks and experiment design for transgenic experiments:

The standard P1-rr4B2 allele, designated as P1-rr for simplicity, of the p1 gene was used in this study (Grotewold et al. 1991). To create a P1-rr stock suitable for use in the Arizona field conditions, an F1 hybrid combining two inbreds (4Co63 and W23), both introgressed with the P1-rr4B2, were crossed to produce a vigorous F1 hybrid. Similarly, the P1-ww stock was an F1 hybrid of P1-ww[A619] and P1-ww[4Co63] inbreds. Both P1-ww alleles carry a single copy of the paramutagenic P1.2 sequence as determined by Southern blot hybridization and neither participate in paramutation (L. Sidorenko, unpublished data).

To produce seed for heritability tests, the transgenic plants were crossed with the P1-ww tester. To verify transgene absence in segregating progeny, DNA was extracted from all herbicide-sensitive plants, and slot blot analysis was done using 32P-labeled GUS probe. To produce seed stocks for secondary paramutagenicity tests P1-rr/P1-ww; TR+/−, plants were crossed with the P1-rr tester and ears exhibiting strong P1-rr′ silencing were selected as a seed source for subsequent generations. To identify nontransgenic P1-rr/P1-rr plants, all herbicide-sensitive plants were screened using hybridization with 32P-labeled GUS probe. Then DNA of all nontransgenic plants was subjected to PCR amplification and genotyping by sequencing to distinguish between P1-rr′/P1-rr and P1-ww/P1-rr siblings. Primers are available upon request.

mop1-1 establishment test:

Details of crosses involved in this test are presented in supplemental Figure 1. To test establishment of paramutation, P1-rr/P1-rr; P1.2b∷GUS/− plants heterozygous for mop1-1/Mop1 were crossed with P1-rr/P1-rr; mop1-1/Mop1 and the resulting progeny were treated with herbicide to eliminate nontransgenic plants. Of the remaining transgenic plants, 1/4 were expected to be mop1-1 homozygous plants, which are informative with respect to the mop1-1 effect on establishment of paramutation. To test P1-rr segregation from these plants, homozygous mop1-1 plants were pollinated with the P1-ww tester and the resulting progeny were grown to evaluate whether P1-rr segregated. We were able to obtain seed from homozygous mop1-1 ears because of the mixed background.

mop1-1 maintenance test:

To assay whether the mop1-1 mutation can reverse P1-rr′ silencing, P1-rr′ was crossed with a standard B′ mop1-1 stock carrying the P1-wr allele (supplemental Figure 2). Two outcrosses to the P1-wr; mop1-1 stock and self-pollination were employed to produce populations in which the effect of homozygous mop1-1 on P1-rr′ expression was assayed (see supplemental Figure 2 for details). Southern blot analysis was used for genotyping P1-rr vs. P1-wr. Genomic DNA was digested with EcoRI and KpnI restriction enzymes, blotted, and probed with fragment 15 (Lechelt et al. 1989), which recognizes 8.0-, 4.0-, and 2.7-kb bands in P1-rr and ∼3- and ∼7-kb bands in P1-wr.

DNA extraction and Southern and slot blot analyses:

Genomic DNA was extracted from leaf tissue and Southern blot analyses were carried out as previously described (Sidorenko et al. 2000; Sidorenko and Peterson 2001). For slot blot analysis, ∼5 μg of genomic DNA was denatured in 1/10 volume of 3 m NaOH for 1 hr at 65°. Denatured DNA was neutralized with an equal volume of 2 m NH4OAC and blotted onto nylon HybondN membrane. Spotted DNA was UV crosslinked to a membrane using a Stratalinker (Biorad) and dried at room temperature. Before hybridization, slot blots were rinsed with 2× SSC and hybridized with 32P-labeled GUS probe as described (Sidorenko et al. 2000) for Southern blots.


P1-rr′ silencing occurs at the transcriptional level:

P1.2b∷GUS transgene-induced silencing of the endogenous P1-rr gene resulted in a 2- to 10-fold reduction in transcript levels, depending on the extent of silencing (Sidorenko and Peterson 2001). This decrease could be associated with lower transcription rates or post-transcriptional degradation of p1 transcripts. While the high meiotic heritability of P1-rr′ is more consistent with transcriptional silencing mechanisms, post-transcriptional gene silencing could also occur through the shared sequences between the downstream copies of the paramutagenic P1.2 fragment and the 3′-UTR of the P1-rr gene (Figure 1C). To address this important question, nuclear run-on analysis was performed using immature pericarp tissues where P1-rr is expressed and the P1-rr′ silencing phenotype is dramatic. Results presented in Figure 2 show that, in a P1-rr′ individual, transcription was ∼5-fold reduced as compared to an individual with the P1-rr allele. This 5-fold reduction is within the range of decreased transcript levels observed in individuals with a similar degree of P1-rr′ silencing (Sidorenko and Peterson 2001), suggesting that the P1-rr′ silencing can be accounted for by the reduction in transcription rate. We also monitored the expression of the biosynthetic A1 and C2 genes that are directly regulated by the P1 protein and saw that they were also reduced to a similar extent in P1-rr′ plants.

Figure 2.—

Nuclear run-on analysis for the P1-rr allele and the P1.2b∷GUS silenced P1-rr′ epialleles in immature pericarp tissue. Nuclei isolated from immature pericarp tissue (18 days after pollination) from homozygous P1-rr and P1-rr′ were used for analysis. The transcription levels for P1-rr and two flavonoid biosynthetic genes directly regulated by p1, a1, and c2 were normalized to the maize Ubiquitin2 gene probe.

P1.2 is sufficient to induce P1-rr′ silencing:

To address whether the P1.2 enhancer is sufficient for paramutation, we produced transgenic constructs in which the Pb fragment was either removed or replaced with heterologous 35S CaMV promoters. To test whether the new constructs could induce paramutation, primary transgenic plants were crossed with nontransgenic P1-rr plants and resulting progeny were assayed for ear pigmentation (Figure 3). Nontransgenic control progeny exhibited no P1-rr silencing (not shown). Similarly, no silencing was observed among transgenic plants containing the negative control constructs that lacked P1-rr sequences (Figure 3; −440 35S∷GUS and −90 35S∷GUS). In contrast, all constructs carrying the P1.2 enhancer sequences induced silencing of the endogenous P1-rr allele, independent of which promoter sequences were present (Figure 3). The frequencies observed were similar to that of the newly produced transgenic events of the positive control P1.2b∷GUS construct. In addition to the high frequency of silencing, these lines also showed similar strengths of silencing; >90% of the silenced plants displayed phenotypes indicating strong P1-rr′ silencing, similar to that shown in Figure 1B. These results demonstrate that sequences contained within the P1.2 fragment are sufficient to establish strong silencing of the homologous P1-rr gene and that no promoter is required.

Figure 3.—

The P1.2 fragment is sufficient to silence the P1-rr gene. (Top) The crossing scheme. Sequences contained within each construct are diagrammed to the left with constant construct components marked above. The Pb promoter and 35S promoter deletions are variable in the constructs and are shown as solid and shaded rectangles, respectively. Silencing efficiency results for transgenic lines containing each construct are shown to the right.

Sequences within P1.2 are sufficient to induce heritable P1-rr′ silencing and secondary paramutation:

To test whether silencing established by transgenes lacking the Pb fragment was heritable, silenced transgenic plants (P1-rr/P1-ww; TR+/−) were crossed with P1-ww tester, which has colorless pericarp and does not participate in paramutation (Figure 4). Examination of the nontransgenic P1-rr/P1-ww progeny class revealed that the positive control (P1.2b∷GUS) and the promoterless (P1.2∷GUS) or promoter replacement (P1.2-44035S∷GUS and P1.2-9035S∷GUS) transgenes all induced highly heritable P1-rr′ silencing in that 83–98% of the progeny remained silenced in the next generation. Previously, we showed that P1-rr′ induced by the P1.2b∷GUS transgene not only was heritably silenced, but also acquired the ability to cause paramutation, which is referred to as secondary paramutation to distinguish it from the initial silencing event, in this case caused by the transgene. To test whether the P1.2 sequences were sufficient for secondary paramutation, strongly silenced P1-rr/P1-ww; TR+/− plants were crossed with the endogenous P1-rr allele and the nontransgenic P1-rr/P1-rr′ individuals were examined for pigment levels to determine if secondary paramutation occurred (Figure 5). All of the events tested from each of the transgenic lines were able to induce a P1-rr′ state that could paramutate a naive P1-rr allele, indicating that the P1.2 enhancer fragment is sufficient to establish P1-rr′ alleles that are both meiotically heritable and capable of inducing secondary paramutation.

Figure 4.—

Heritability of silencing induced by promoter replacement constructs. (Top) The crossing scheme and predicted genotypes. In all cases, the female parents used in the crosses demonstrated strong P1-rr′ silencing phenotypes. Sequences contained within each construct are diagrammed to the left with constant construct components marked above. The Pb promoter and 35S promoter deletions are variable in the constructs and are shown as solid and shaded rectangles, respectively. Silencing efficiency results for transgenic lines containing each construct are shown on the right.

Figure 5.—

Secondary paramutagenicity of promoter replacement constructs. The crossing scheme and predicted genotypes are indicated on the top. In all cases, the female parents used in the crosses demonstrated strong P1-rr′ silencing phenotypes. Sequences contained within each construct are diagrammed to the left with constant construct components marked above. The Pb promoter and 35S promoter deletions are variable in the constructs and are shown as solid and shaded rectangles, respectively. Silencing efficiency results for transgenic lines containing each construct are shown on the right.

Establishment of P1-rr′ silencing requires MOP1:

Previous studies have shown that MOP1 is absolutely required for the establishment, the transmission of the silencing from paramutagenic to paramutable alleles in trans, of b1, pl1, and r1 paramutation (Dorweiler et al. 2000). To test whether p1 paramutation, which is induced by transgenes, also requires MOP1, we used a series of crosses to combine P1-rr with P1-rr′ and the P1.2b∷GUS transgene in a homozygous mop1 mutant background, mop1-1 (supplemental Figure 1). If MOP1 is required for paramutation, we would expect that P1-rr would not be silenced in the mop1-1 mutant, producing ears with red pericarps. As shown in Table 1, that expectation was met in the majority (20/23) of homozygous mop1-1 plants from two different lineages. In contrast, in the majority (64/69) of the mop1-1/Mop1 and/or Mop1/Mop1 plants, strong silencing was observed. There were a few weakly silenced individuals (orange pericarps) that were homozygous for mop1-1 or carried at least one functional Mop1 allele (Table 1). The presence of this class in the fully recessive mop1-1 homozygous plants suggests that at a low frequency the transgene may be able to mediate weak silencing even in the absence of MOP1. However, the lack of strongly silenced P1-rr′ individuals in mop1-1 homozygous plants demonstrates that the establishment of p1 paramutation is disrupted in the absence of MOP1.

View this table:

The mop1-1 mutation prevents establishment of paramutation at P1-rr

We next addressed whether the apparent P1-rr allele would segregate phenotypically unchanged from the mop1-1 homozygous plants after it was exposed to paramutagenic P1.2b∷GUS and P1-rr′. For this mop1-1 homozygous plants were crossed with the P1-ww allele. P1-ww is colorless, does not participate in paramutation, and enables easy scoring of the P1-rr vs. P1-rr′ phenotype in heterozygous ears (supplemental Figure 1). If P1-rr were heritably changed to P1-rr′, a reduction in red ears and an excess of patterned ears would be expected. If P1-rr escaped paramutation in mop1-1 homozygous plants, then equal ratios of red (P1-rr) and patterned (P1-rr′) ears should be observed among nontransgenic progeny. Red ears were observed in testcross progeny of both transgenic events, supporting our conclusion that P1-rr can escape paramutation in mop1-1 homozygotes. The expected 50% frequency was met with the P147-37 event in which 20 of 44 plants had ears with red pericarps (Table 2). The P147-9 event had more red ears than expected (41 of 69; Table 2). The identical gene structures of P1-rr and P1-rr′ do not allow us to distinguish between skewed segregation and possible reversion of P1-rr′ to P1-rr that is occasionally observed in P1-rr′ stocks (Sidorenko and Peterson 2001). In either case, the increased proportion of red ears does not complicate our interpretation because, if P1-rr had been paramutated, the mop1-1/mop1-1 plants would not have had red pericarp and an excess of light ears would be expected, which was not observed. In summary, these results demonstrate that P1-rr can segregate unchanged from homozygous mop1-1, demonstrating that MOP1 is required for efficient establishment of P1-rr′ silencing.

View this table:

P1-rr segregates phenotypically unchanged after being exposed to paramutagenic P1.2b∷GUS and P1-rr′ in the mop1-1 homozygote

P1-rr′ silencing is not immediately released by the mop1-1 mutation:

In addition to preventing the establishment of paramutation, when the mop1-1 mutation is combined with b1 and pl1 alleles, the maintenance of silencing previously established by paramutation is immediately reversed (Dorweiler et al. 2000), resulting in increased transcription of the previously silenced alleles. Importantly, the mop1-1 mutation did not increase the expression of several b1 or pl1 alleles that do not undergo paramutation, suggesting the MOP1 protein is not a general negative regulator at these loci (Dorweiler et al. 2000). We assessed the effect of mop1-1 on the maintenance of P1-rr′ silencing by crossing P1-rr′ with our standard Bmop1-1 stock, which carries the P1-wr allele that lacks pericarp pigmentation (Figure 1A) and does not undergo paramutation (Sidorenko and Peterson 2001) (materials and methods, supplemental Figure 2).

Two p1 genotypes are expected to segregate at equal frequencies from the crosses described in detail in supplemental Figure 2: P1-rr/P1-wr and P1-wr/P1-wr. If the absence of MOP1 releases P1-rr′ silencing, then the mop1-1/mop1-1 ears carrying P1-rr/P1-wr should have red or orange pericarp. This was not the case, as no mop1-1/mop1-1 ears with darkly pigmented pericarps were observed in the first generation (Table 3; Figure 6, A and B). This result contrasted with the results at b1 and pl1, in which within the first generation the silencing was reversed to the highest levels by the absence of MOP1 in essentially all plants. Precedence for the need for prolonged exposure to mop1-1 mutant backgrounds for reactivation comes from experiments with reactivation of MuDR transposons. High frequencies of MuDR transposon reactivation were achieved only after six generations in the homozygous mop1-1 background (Woodhouse et al. 2006).

Figure 6.—

Ear phenotypes of mop1-1/Mop1 and mop1-1 plants. Three representative ears are shown for each group to demonstrate the range of variation observed in pericarp phlobaphene pigmentation in the mop1-1 maintenance test. (A) The P1-rr/P1-wr; mop1-1/Mop1 plants have a codominant phenotype; dark red cob is specified by P1-wr and light stripes and pigmentation at the silk scar are contributed by P1-rr′. (B) After one generation of exposure to homozygous mop1-1, the P1-rr/P1-wr ears remain lightly pigmented. (C) After two generations of exposure to mop1-1/mop1-1, the P1-rr/P1-wr ears exhibit sporadic increases in pericarp pigmentation: light orange blush throughout the pericarp and increased pigmentation near the silk scar. (D) After three generations of exposure to homozygous mop1-1, dark orange and red pericarp pigmentation can be observed in P1-rr/P1-wr. Ears of the P1-rr/P1-rr′ ears (not shown) were indistinguishable from P1-rr/P1-wr. (E) P1-wr/P1-wr ears in mop1-1/Mop1 background have colorless pericarp and red cob typical for P1-wr phenotype. (F) After one generation of exposure of both P1-wr alleles to at least one generation of mop1-1/mop1-1, some P1-wr/P1-wr ears exhibit increased pericarp pigment that we term “blushed.” (G) After two generations of exposure to mop1-1/mop1-1, further increases in P1-wr pericarp pigmentation can be observed.

View this table:

Partial P1-rr′ reactivation is observed after two generations of exposure to the mop1-1 mutation

Although all mop1-1/mop1-1 ears were lightly pigmented, an unusual blushed ear class was seen in some homozygous mop1-1 plants while only the typical patterned or colorless ears characteristic of P1-rr/P1-wr were observed in the control ears heterozygous for mop1-1 (Table 3). The presence of the unusual blushed ear class in homozygous mop1-1 plants suggested that there might be an occasional effect of the loss of MOP1 on either P1-rr′ or P1-wr. Subsequent crosses described below were carried out to assess the effect on expression of both alleles by multiple generations in the homozygous mop1-1 mutant background.

Prolonged exposure to the mop1-1 mutation gradually increases P1-rr′ and P1-wr pericarp pigmentation:

To test whether an extended exposure to mop1-1 would increase the expression of P1-rr′, we pollinated P1-rr′/P1-wr; mop1-1/mop1-1 plants with pollen from the P1-wr/P1-wr; mop1-1/Mop1 tester (supplemental Figure 2). Examination of the resulting mop1-1/mop1-1 progeny revealed that there was a reduction in the number of plants with strongly silenced colorless and patterned pericarps and that three plants had orange pericarp pigmentation (Table 3, Figure 6C). The three individuals with orange pericarps were genotyped and shown to carry P1-rr′/P1-wr. On the basis of pedigree, the P1-rr′ allele in these individuals came from a female parent homozygous for mop1-1 in the previous generation. Thus, in the three plants with increased pigmentation, P1-rr′ was exposed to mop1-1 for two consecutive generations. To test whether pericarp pigmentation would continue to increase in subsequent generations in the mop1-1 background, the P1-rr/P1-wr; mop1-1/mop1-1 plants were self-pollinated (supplemental Figure 2). The resulting progeny were homozygous for mop1-1 and carried P1-rr′ and P1-wr that had been in a homozygous mop1-1 background for three and two consecutive generations, respectively. As shown in Table 4, there was further reduction in silencing; 7 of 28 ears had red pericarp pigmentation, and the remainder had partial silencing phenotypes (orange, patterned, and blushed) with no fully silenced (colorless) individuals observed. Examples of the phenotypes are shown in Figure 6, D and G. Southern blot analysis revealed that two P1-rr/P1-rr′, four P1-rr/P1-wr, and one P1-wr/P1-wr plant had red pericarp. Thus, the silencing of both the paramutagenic P1-rr′ allele and the P1-wr allele that does not participate in paramutation could be fully or partially released through prolonged exposure to the mop1-1 mutation.

View this table:

Homozygous mop1-1 exposure leads to increased pigmentation after multiple generations of exposure

Tissue-specific pigmentation pattern of the P1-wr allele undergoes modification in the presence of the homozygous mop1-1 mutation:

The upregulation of P1-wr in some mop1-1 homozygous plants is consistent with the hypothesis that MOP1 is required for the silencing of P1-wr in the pericarp, but not in the cob. We suspected that we had not observed this previously because our P1-wr mop1-1 stock is routinely propagated using heterozygous mop1-1/Mop1 ears and pollen from homozygous mop1-1/mop1-1 plants to circumvent the deleterious effects of the mop1-1 mutation on ear fertility in inbred backgrounds. As the pericarp is a maternal tissue, none of our ears would be homozygous for mop1-1. However, other more complicated scenarios could be imagined, given that we are studying allele interactions. Thus, we conducted a separate control experiment in which effects of mop1-1 on P1-wr could be monitored in the absence of any prior history with P1-rr′. For this we self-pollinated ears of P1-wr/P1-wr; mop1-1/mop1-1 or P1-wr/P1-wr; mop1-1/Mop1 (not shown). Yield was poor with the homozygous mop1-1 ears, which was not unexpected, given past experience with mop1-1 in inbred backgrounds. Examination of the resulting 19 mop1-1 ears revealed that 14 individuals (74%) exhibited blushed pericarp pigmentation similar to that shown in Figure 6F, while only colorless pericarps were observed for the 57 mop1-1/Mop1 controls. The extreme reduction in vigor of the homozygous mop1-1 plants prevented the examination of pigment levels in additional generations and testing heritability of P1-wr upregulation. The P1-wr phenotype is extremely stable; no spontaneous increases in pericarp pigmentation have been observed in many thousands of wild-type P1-wr plants examined (Chopra et al. 1998). Thus, the modest upregulation of P1-wr in the mop1-1 mutant background after one or two generations of exposure is significant and suggests that the tissue-specific silencing of the P1-wr allele involves MOP1.


These results provide further insights into the mechanism of p1 paramutation. The demonstration that P1-rr′ silencing occurs at the transcriptional level, and that establishing the silencing requires MOP1, a RNA-dependent RNA polymerase, strongly suggests that p1 paramutation is mediated through RNA-directed chromatin alterations. The loss of MOP1 completely prevents establishment of silencing, but once established in wild-type backgrounds, it is only after several consecutive generations in the absence of MOP1 that reactivation of a previously silenced allele is observed. Our transgenic experiments refined the sequence requirements for p1 paramutation by demonstrating that no promoter sequences are required and that the P1.2 enhancer fragment is not only required for paramutation—it is sufficient. Additionally, our result that MOP1 contributes to maintaining the tissue-specific silencing pattern of the P1-wr allele, which does not undergo paramutation, expands the number of processes regulated by RNA-mediated transcriptional silencing mechanisms.

p1 paramutation involves transcriptional silencing:

Previous studies with multiple systems suggest that paramutation can be mediated at the transcriptional or post-transcriptional levels, depending on the system. Our results with p1 paramutation are most similar to the results reported for the b1 gene of maize (Patterson et al. 1993) in that the fold reduction in transcript levels can be accounted for by reductions in transcription. Two other paramutation examples that show significant reduction in transcription include a transgenic locus in Arabidopsis (Mittelsten-Scheid et al. 2003) and the pl1 locus in maize (Hollick et al. 2000; Hollick and Chandler 2001). In addition, at pl1 post-transcriptional changes have also been documented, but as the key sequences mediating paramutation have not yet been defined, mechanistic interpretations are difficult. The best-characterized example of paramutation in a mammalian system is hypothesized to occur at the post-transcriptional level as the phenomenon can be partially recapitulated by injecting miRNAs into blastocysts (Rassoulzadegan et al. 2006). Further experiments will be required to determine the mechanisms through which transcription is altered at p1, but previously described correlations with increased DNA methylation (Sidorenko and Peterson 2001) combined with the transcriptional silencing shown here make alterations in chromatin structure an attractive model.

The P1.2 enhancer sequence is sufficient to establish and maintain transcriptional silencing associated with p1 paramutation:

Our transgenic experiments demonstrated that sequences within the basal P1-rr promoter and 5′-UTR were not required for paramutation. In fact, no promoter is required at all; the P1.2 enhancer fragment is sufficient to establish P1-rr′ that is heritable and can induce secondary paramutation. The observation that the basal P1-rr promoter is not required within the transgene for p1 paramutation is important because it eliminates a requirement for the highly repetitive 168-bp transposon-related sequence there (D. Lisch and F. Zhang, personal communication). Although we showed that the repetitive sequences within Pb are not required for paramutation, the role of the repeats in p1 paramutation remains under investigation. This is because the P1.2 fragment required for paramutation is repeated within the P1-rr locus and the 719-bp subfragment of P1.2 is moderately repeated elsewhere in the genome (∼10 copies in addition to sequences present at the p1 locus). The 361 bp of the moderately repetitive 719-bp subfragment of P1.2 are transcribed as a part of the 3′-UTR of P1-rr, while another 322 bp share >95% sequence similarity to transcripts originating from other genomic locations (Sidorenko and Peterson 2001). The minimal sequences required for paramutation have been identified for one other maize locus, b1. At b1, the required sequences are located ∼100 kb upstream of the b1 gene within a 6-kb fragment containing an 853-bp sequence that is repeated in tandem seven times in alleles that participate in paramutation and only once in alleles that do not (Stam et al. 2002). This tandemly repeated 853-bp sequence required for paramutation is unique to this locus as it is found nowhere else in the maize genome. In nuclear run-on experiments transcription was detected for both DNA strands of the repeats while no protein-coding capacity could be predicted using sequence analysis (Alleman et al. 2006). Tandem repeats are also required for paramutation at the r1 locus, but there are major differences when compared to b1. At r1 the tandem repeats span the r1-coding regions and are of unknown length (Eggleston et al. 1995). Recombination mapping generated a series of alleles with differing numbers of tandem repeats (Kermicle et al. 1995; Kermicle 1996; Panavas et al. 1999) and their study revealed that no specific region within the repeat is required: it is the number of repeats that matters with the strength of paramutation correlating with the repeat numbers. Thus, an attractive hypothesis is that some aspect of the repeat structure at P1-rr that encompasses the P1.2 fragment (Figure 1) contributes to paramutation. Further experiments to examine the role of the repeats and transcribed sequences, both within the transgenes inducing paramutation as well as the endogenous alleles responding, should help to refine our understanding of the sequence requirements for p1 paramutation.

p1 paramutation has specific requirements for MOP1:

Our experiments with the mop1-1 mutation provide the first insights into the proteins and pathways mediating p1 paramutation. Similar to paramutation at the b1, pl1, and r1 loci (Dorweiler et al. 2000), the establishment of p1 paramutation depends on the mop1-encoded RNA-dependent RNA polymerase. When a previously silenced allele at b1 or pl1 is crossed into a mop1 mutant background, the silencing is immediately relieved with full expression restored (Dorweiler et al. 2000). In contrast, P1-rr′ silencing is released gradually; only after multiple consecutive generations in the mop1-1 mutant background were strong increases in P1-rr′ pericarp pigmentation observed. The slow, multi-generational reactivation of P1-rr′ is reminiscent of the reactivation of the MuDR transposon by mop1-1: excision was only 3% in the first generation (Lisch et al. 2002), but increased to 36% after six generations in the mop1-1 background (Woodhouse et al. 2006). An additional example where mop1-1 was not required to maintain paramutation-like silencing was reported for the maize lpa1 (low phytic acid1) locus (Pilu et al. 2008). Our findings suggest that the RNA-mediated pathway involving mop1 is absolutely required to establish p1 paramutation, but that other mechanisms can at least partially substitute to maintain silencing of previously silent paramutated alleles and silenced transposons. This is in contrast to other systems like b1 and pl1, where MOP1 is required both to establish silencing and to efficiently maintain that silencing. Molecular evidence suggests that mop1-1 is a null allele and phylogenetic analysis indicates that mop1 encodes the only maize RDR in the rdr2 clade (Alleman et al. 2006). It is possible that a distinct maintenance pathway involving the other rdr genes present in the maize genome (http://www.chromdb.org) plays a role in maintenance of silencing at p1. Testing additional mutations that affect paramutation at b1 and pl1 for their effect on p1 paramutation, together with detailed chromatin and DNA methylation studies within the key sequences mediating b1 and p1 paramutation in the presence and absence of MOP1, may shed light on the potential mechanisms and reasons for the differences at each locus.

RNA mechanisms participate in the tissue-specific silencing of P1-wr:

It is intriguing that tissue-specific silencing of the P1-wr allele, which does not participate in paramutation, is relieved in the mop1-1 background. This contrasts with studies with multiple b1 and pl1 alleles that do not participate in paramutation, which are not upregulated by the mop1-1 mutation (Dorweiler et al. 2000). In most other genetic backgrounds, including multiple inbred stocks, P1-wr has an extremely stable phenotype producing colorless pericarp and red cob (Chopra et al. 1998). The P1-wr allele has a complex structure with approximately six gene copies arranged as tandem repeats (Chopra et al. 1998) and extensive DNA methylation throughout P1-wr is observed (Chopra et al. 1998, 2003). These results, combined with the observation from transgenic experiments that the P1-wr promoter fused to P1-wr cDNA can function in both pericarp and cob tissues (Cocciolone et al. 2001), suggest epigenetic regulation of P1-wr. The shared involvement of MOP1 in paramutation and tissue-specific silencing suggests that some aspect of P1-wr structure, potentially the tandem repeats, targets it for heterochromatic silencing by the RNA-mediated pathway involving MOP1. Another potential component in that pathway might be ufo1 (unstable factor for orange1) because P1-wr expression is upregulated in the Ufo1 mutant background, resulting in fully pigmented ears (Chopra et al. 2003).

The observation that the P1-wr locus does not participate in paramutation with P1.2b∷GUS or P1-rr′ (Sidorenko and Peterson 2001), in spite of having tandem repeats and being silenced epigenetically by a MOP1-dependent pathway, demonstrates that simply having tandem repeats and being epigenetically silenced is not sufficient for paramutation. Further experiments to examine the role of other mutants on P1-rr′ and P1-wr expression and whether or not siRNAs are produced should provide further mechanistic information on the requirements for paramutation.


We are grateful to the Iowa State Transformation Facility for producing the transgenic plants on a fee-for-service basis. We also thank Joshua Farr for technical assistance and field manager Carl Schmalzel for excellent plant care. We are grateful to Thomas Peterson (Iowa State University) for his help with transgenic experiments and his continued support and encouragement. Research was funded by U. S. Department of Agriculture grants 01740 and 35301-13243 to V.C.


  • Communicating editor: V. Sundaresan

  • Received August 19, 2008.
  • Accepted October 9, 2008.


View Abstract